Pardoned Offenders in Canada: A Statistical Analysis 2000-02

By Suzanne Wallace-Capretta, M.A.
Department of the Solicitor General Canada
March 2000

The views expressed are those of the author and are not necessarily those of the Department of the Solicitor General of Canada.


Without the support and cooperation of a number of people this report would not have been possible. The assistance of Tanya Rugge in collecting the data for the pilot study as well as coordinating the data collection of the present evaluation is very much appreciated. Thanks to the research contractors Ian Broom, Malorie Lapointe, Kevin Nunes and Paul Verbrugge for collecting the data in a timely manner. My appreciation is also extended to Amy Yuile for her help with the literature search. As well, I would like to thank Eleanor King for her cooperation and George Sieniecky and Daniel Quesnel for helping us locate the records. Thanks to Mia Dauvergne, Robert Cormier, Mary Campbell, Karl Hanson, Cliff Yumansky, Ghislaine Charlebois, Pierre Couturier, Sheila Ouellette and Colette Galipeau for comments on an earlier draft. Finally, thanks are extended to Jim Bonta and Richard Zubrycki for their helpful advise and support.


Although there have been over 234,779 pardons granted between 1970 and 1998, little is known about the characteristics of ex-offenders who were granted a pardon. During this same period, 6,046 pardons were revoked or ceased to be in effect and 2,333 pardon applications were denied. In order to understand more about pardoned offenders, three studies were conducted. Study 1 examined the characteristics of a sample of pardon applicants. Pardoned offenders

(n = 1,128) were compared with those who were denied (n = 330). The results of Study 1 showed that the denied sample had more serious and lengthier criminal histories than the granted sample. The second study was concerned only with those offenders who were granted a pardon. Two samples were evaluated: the first, a sample of successfully granted pardons (n = 603) and the second consisted of individuals whose pardons were subsequently revoked or ceased to be in effect (n = 525). Study 2 found that individuals whose pardons were revoked had more prior offences leading to more prior convictions than those who were not revoked. Also, pardons were most commonly revoked for property and liquor/traffic offences. Finally, Study 3 examined a sub-sample of sex offenders whose pardons were revoked or ceased to be in effect. The findings of the third study indicated that approximately six per cent of the revocations were the result of a sex offence. Extrapolations were made from the sample figures to provide estimates of sex offending in the pardoned population.


The concepts of clemency and pardon have existed, as long as there have been laws. The gods of ancient civilizations were believed to possess the power to issue clemency or pardon that was dispensed by the King or state leader. Through the centuries the power to exercise clemency and pardon shifted repeatedly between the Church and the Monarch. The general purpose of the Royal Prerogative of Mercy in England, from its beginnings, was to ".compensate for the strictness of the criminal law" (Couillard, 1985). Although the intent of clemency and pardon was to ensure justice and allow for mercy, it was at times exploited by requiring monetary compensation from the offender or to recruit laborers for the King's plantations or navy.

Since clemency and pardon were the only form of reprieve until the creation of legal appeals and sentence remission, several additional purposes of clemency and pardon have emerged. The first purpose of clemency or pardon was to undo a wrong. For example, given that any system of justice is not perfect, a pardon provided recognition that an individual had been wrongfully convicted. A pardon could also be a form of mercy. That is, the decision to convict the individual was correct but the full penalty was not exacted for reasons of undue hardship or excessive inequity or injustice. Thirdly, a pardon could be perceived as a form of forgiveness. An individual who was convicted of a crime and fulfilled his/her sentence often encountered unnecessary difficulties thereafter (Smith, 1983). If this person demonstrated that he/she had become a responsible member of society then forgiveness would come in the form of sealing or expunging the record and fully restoring civil rights and liberties. In this form, the intent of a pardon was to reduce any negative stigma associated with conviction. These general purposes of clemency and pardon appear to exist within the criminal justice systems of many countries under a variety of names (Burton, Travis, & Cullen, 1988; Couillard, 1985; Smith, 1983; Solicitor General Canada, 1981).

In Canada the first two purposes of clemency and pardon are embodied within the Royal Prerogative of Mercy and the third purpose, forgiveness, in the Criminal Records Act (CRA). Stemming from the British Royal Prerogative of Mercy in which the discretionary powers of clemency and pardon are held by the Monarch, the Governor General of Canada has the power to exercise clemency and pardon by virtue of the Letters Patent. The Governor in Council also has the authority to grant clemency and pardon as set out in sections 748 and 749 of the Criminal Code of Canada. Under exceptional circumstances of injustice or hardship and when no other remedy is available, the Governor General or the Governor in Council may utilize their discretionary powers. These powers include several options. For example, a free pardon may be granted if the conviction was in error. Another available option is a conditional pardon prior to eligibility under the CRA or the Corrections and Conditional Release Act. Additional options are remission of sentence, remission of fine, relief from prohibitions or respite (interrupting a sentence). Since this power to right a wrong or grant mercy is invoked under extreme circumstance, it has been exercised infrequently. Combining all cases involving a free or conditional pardon or some form of remission, relief or respite the Royal Prerogative of Mercy has been exercised 159 times since 1980 (National Parole Board, 1999).

Unlike the discretionary power of the Royal Prerogative of Mercy, which applies to exceptional cases and is guided by general principles, detailed legislation has been set out in order to govern the granting of pardons for the purpose of forgiving past wrongdoings (violations of Federal acts and regulations). Originally enacted in 1970, the CRA provided that an individual was eligible to submit an application for a pardon, after fulfilling the requirements of his/her sentence, if he/she had been of "good conduct" for a prescribed length of time. Good conduct for the purposes of the CRA is defined as "a conviction-free period, with no suspicion or allegation of criminal behaviour" (National Parole Board, 1999). A pardon under the CRA enabled an offender who had demonstrated that he/she had successfully reintegrated into society to reduce the legal, economic and social limitations associated with possessing a criminal record. A pardon effectively seals an offender's record requiring that it be held separately. It does not permanently remove or destroy the record. A pardoned record held in federal custody cannot be revealed without the approval of the Solicitor General of Canada. However, the National Parole Board (NPB) may revoke a pardon if the individual has been convicted of further offences, is no longer of "good conduct", or falsified information was presented in order to obtain the pardon.

In accordance with the CRA (1970), pardon applications were received by the Solicitor General of Canada. The Minister in turn referred applications to the NPB for review. The NPB would then conduct an investigation into the applicant's behaviour since the time of the conviction. In most cases, this consisted of a Royal Canadian Mounted Police (RCMP) field investigation that included meeting the applicant, and interviewing neighbours, past and present employers and others who could comment on the applicant's behaviour. The NPB would report the findings of its investigation to the Solicitor General with a recommendation to either grant or not grant a pardon. Subsequently, the Solicitor General Canada was required to refer any recommendations to grant a pardon to the Governor in Council. This complicated process demanded a high level of resources and even then, often resulted in lengthy delays.

In order to reduce the workload demands and produce a more effective and efficient process, amendments were made to streamline the CRA in July of 1992. The authority to grant or revoke a pardon was assigned to the NPB circumventing the previously long and arduous process. Table 1 outlines some of the key changes. For example, prior to 1992 all discharges (absolute and conditional) required a pardon be granted. The amended CRA does not require a pardon for a discharge because, while it implies guilt, no conviction is registered. Instead, a discharge is automatically removed from the offender's record after a designated period of time. In addition, the 1992 amendments made it possible to issue a pardon for summary offences following completion of the sentence and the required conviction free period. Changes were also made to the pardon revocation process. Consequently, a pardon would automatically cease to be in effect as a result of a conviction for a new indictable offence (also referred to as a cessation). In the case of a subsequent summary conviction, the NPB maintained the discretionary power to revoke the pardon.

Table 1. Pardon Eligibility Based on Conviction Free Waiting Periods
Type of offence/disposition CRA (1970) CRA Amendments (1992)
Conviction 5 years - grant
(following sentence completion)
No change - grant
Conditional Discharge 3 years - grant
(after completing conditions)
Pardon not necessary
(automatically removed 3 years after court decision)
Absolute Discharge 3 years - grant
(from court date)
Pardon not necessary
(automatically removed 1 year after court decision)
Conviction 2 years - grant
(following sentence completion)
3 years - issue
(after sentence completion)
Conditional Discharge 1 year - grant
(after completing conditions)
Pardon not necessary
(automatically removed 3 years after court decision)
Absolute Discharge 1 year - grant
(from court date)
Pardon not necessary
(automatically removed 1 year after court decision)

Note: In the case of a summary offence the CRA amendments dictate that a pardon can be issued following the designated conviction free period and verification of no further criminal activity.

Following the adoption of the 1992 legislative changes, policy changes were made in 1996 by the NPB as part of the Federal Government program review exercise to streamline government services. After consulting with several criminal justice partners, the NPB implemented policy changes that were expected to further improve the pardon application process (Solicitor General Canada, 1997). These changes were as follows:

  1. RCMP field investigations would no longer be required.
  2. The applicant would be required to submit a copy of his/her criminal record and a local police records check of the past 5 years, issued within the previous six months.
  3. If the time between sentence completion and the date of application was less than five years, confirmation that the offence was proceeded with summarily would be required and certification that any fines, surcharges, restitution or other forms of compensation had been paid.
  4. A $50 administration fee would be required for processing each application.

Noticeable differences in the number of pardons granted and issued have resulted from the amendments to the CRA (1992) and to a lesser degree the policy changes implemented by the National Parole Board. Figure 1 presents the number of pardons granted and issued between 1987/88 and 1997/98. Although 5,748 pardons were issued (i.e., non-discretionary) for the first time in 1992/93 as a consequence of the CRA amendments, the number of pardons granted in that same year also doubled from previous years. Following the NPB policy changes of 1996 a decrease in the number of pardons granted and issued occurred in 1997/98. The requirement for the applicant to contact the local police and the introduction of an application fee of $50 may have contributed to the decrease.

Figure 1: Pardons granted and issued (1987/88-1997/98)
Image Description

Figure 1 displays a bar chart of pardons granted and issued for each fiscal year from 1987/88 to 1997/98 . The first five years fluctuated between 8,374 and 11,360 pardons granted. In fiscal year 1992/1993 there was a spike in 29,205 pardons granted and issued. The number of pardons granted and issued varied between 15,401 and 23,895 in the subsequent four years and then fell to 7,633 in 97/98.

With respect to revocations, the pattern is similar to that found among the pardons. Steady increases can be seen in the number of revocations beginning in 1992 followed by a decrease after the policy changes in 1996. As illustrated in Figure 2, pardon cessations, which commenced in 1993/94, followed a similar pattern. That is, the number of cessations increased annually until the implementation of the 1996 policy changes after which it declined. The number of pardon denials however, has remained fairly constant over the 10 years indicating that they were less affected by both the legislative and policy changes.

Figure 2: Revocations, cessations, denials (1987/88-1997/98)
Image Description

Figure 2 is a graphical display of revocations, cessations, the total revoked and ceased, and denials for each fiscal year from 1987/88 to 1997/98. The number of denials remained steady for all years at under 200. The total number of revocations showed an increase from 200 in 1992/93 to nearly 800 in 1993/94 and these numbers continued to increase to over 1,200 in 1997/97 before falling to approximately 700 in the last recorded year. Cessations and revocations also mirrored this trend but at a lower number.

Between 1970 and 1998, over 234,779 pardons were granted and issued. However, little information has been recorded about the characteristics of offenders who were granted/issued or denied a pardon, or had their pardon revoked/ceased to be in effect. During this same period, 6,046 pardons resulted in a revocation or cessation yielding an overall revocation rate of 2.6%. Even smaller is the frequency of pardon denials. Compared to pardons granted, very few are denied in any given year (ranging from 68 to 228; NPB, 1998).

The Present Studies

Comparative information was called for to support policy development on the issue of making available pardoned sexual offence records for screening purposes. It was already possible to review active criminal history records through the National Screening System (1996), but consideration was being given to opening some pardoned records for the same purpose.Footnote 1 A review of the literature found very little research examining pardon decisions in Canada and the offenders affected by these decisions. In order to address this gap in knowledge, three studies were conducted. Study 1 compared 1,128 pardoned offenders with 330 offenders who were denied a pardon. This comparison permitted an assessment of the appropriateness of decisions to grant pardons. The second study was concerned with the long-term outcome of offenders who were granted a pardon. Study 2 involved two samples. The first sample consisted of 603 offenders for whom there was no documented instance of further criminal activity. The second consisted of 525 individuals whose pardons were subsequently revoked or ceased to be in effect. The criminal histories of the two groups were compared. As well, the types of offences resulting in revocations or cessations were examined. Finally, Study 3 examined a sub-sample of sex offenders whose pardons were revoked or ceased to have effect. The serious consequences of sex offences are of substantial concern to the public. The purpose of the third study was to determine the reoffence rate of pardoned sex offenders. The data for all three studies were obtained from reviews of NPB files carried out in February, 1999.

Study 1: The Granting of Pardons

Study 1 describes some of the factors related to the granting of pardons. A sample of pardon applications was collected which included 1,128 granted pardons and 330 denied pardons. The years 1988 to 1992 were chosen for study for a number of reasons. First, these years were selected so that all of the decisions to grant or deny occurred prior to the implementation of the 1992 CRA amendments. As a result, the sample does not contain pardons "issued" for summary convictions. In this way, we could examine NPB decisions to grant or deny a pardon for both summary and indictable offences while excluding non-discretionary decisions (issued pardons). Second, anticipating the requirements for Study 2, we selected a sample that provided a follow-up period of reasonable length to examine outcome.

A limitation to the methodology was that pardon applications were randomly sampled according to calendar year while the NPB organizes the files by fiscal year. The NPB denied 446 pardon applications between April 1, 1988 and March 31, 1993. Considering the relatively small number of denials, we attempted to collect as many denials occurring between 1988 and 1992 as could be located. Due to the large number of granted pardons, random samples were drawn from each year. Another important restriction was that a typical file contained limited information. The information available included personal demographics such as gender, employment, date of birth, province of residence, and criminal history information taken from the RCMP Finger Print Service Record. Also available for many cases were the RCMP investigation report and a record of the decision by the NPB.


An analysis of the personal demographic data showed few noteworthy differences between the pardoned and denied groups. Females were more likely to be granted a pardon than males (89% vs. 76%, X Footnote 2 = 16.6, p < .001), whereas, no statistically significant differences were found in terms of employment (71% vs. 67%) or age at the time of application (average of 34 years for both groups). Not unexpectedly, the majority of the pardon applications (90%) were from the four most populous provinces: Quebec (29.8%), Ontario (26.1%), British Columbia (18.2%) and Alberta (16.1%). Examining the rates at which pardons were granted and denied in these four provinces two significant differences were found. First, more applications received from British Columbia were denied compared to the other three provinces combined (39.1% vs. 14.6%, X = 85.5, p < .001). Second, the opposite was true for applications received from Quebec, with fewer being denied in Quebec than from Ontario, British Columbia and Alberta combined (18.5% vs. 37.3%, X = 36.6, p < .001). In terms of the amount of time taken to reach a decision, on average it took 738 days (range 716 - 783 days) to deny a pardon application and 457 days (range 426 - 478 days) to grant a pardon (t = 15.3, df = 1,456, p < .001). This difference was significant within all four provinces but no differences were found between them.

The two groups were found to have significantly different offence histories. The denied applicants were convicted of their first offence at an earlier age, an average of 22 years versus 24 years for the granted group (t= 4.9, df = 1,452, p < .001). The analysis also revealed that denied applicants were more than three times as likely to have been previously denied a pardon than those in the granted sample (10.3% vs. 3.1%, X = 29.4, p < .001). Furthermore, differences were evident when examining the most serious offence prior to their application. The most serious offence was determined by the severity of the sanction received and by the severity of the offence committed. Assuming the least restrictive principle was applied at sentencing, a custodial sentence would be coded as the most serious disposition and a discharge as the least serious. When a disposition deemed most serious appeared for more than one offence (e.g., 6 months incarceration for an assault and 6 months incarceration for a break and enter) the more serious offence was coded. For example, crimes against a person range from murder, most serious, to uttering threats, least serious. As well, crimes against a person were coded as more serious than narcotics, liquor and traffic violations (e.g., impaired driving) and property offences, in that order. In general, individuals whose applications were denied had been convicted of more crimes against a person and fewer property crimes than granted applications (Table 2). As expected, the sanctions imposed for the most serious offence were harsher for the denied group, 61.2% were imprisoned while only 27.9% of the granted cases served a custodial sentence. Reviewing the entire offence history, not just the most serious crimes, we found that the denied applicants were more likely to have been convicted of a previous sexual offence and/or a non-sexually violent offence.

Table 2. Offence History of Granted and Denied Pardon Applications
Variable Granted Denied t/X p
  (n = 1,128) (n = 330)    
# of prior convictions (mean) 2.2 4.1 10.0 ***
# of prior offences (mean) 3.2 6.2 7.0 ***
Any prior sexual offences (%) 3.8 8.8 13.5 ***
Any prior non-sexually violent offences (%) 16.0 33.3 47.8 ***
Most Serious Offence (%)
Property 42.0 33.9 6.9 **
Liquor/Traffic 24.6 20.0 3.1 n.s.
Narcotics 15.3 18.8 2.5 n.s.
Person 9.0 14.5 8.4 **
Sexual 2.5 6.4 11.8 ***
Other 6.5 6.4 0.0 n.s.

Note: *** = p < .001, ** = p < .01, n.s. = nonsignificant

Each file documented the reasons for the pardon denial. Various explanations were provided, most of them related to a failure to meet the requirement of "good conduct". Violations of "good conduct" accounted for over 36% (121/330) of the denied applications. Other, more specific reasons given for denial were continued association with criminals, a police incident resulting in a warning, being charged or arrested although not convicted and suspicions of illegal activity (under investigation). A further review of the decisions to deny revealed that substance abuse played a role in the decision 87.4% of the time (180/206).


Study 1 provided a brief analysis of some of the factors related to the granting of a pardon. From the analysis it would appear that the granted sample had fewer factors related to recidivism based on our knowledge of the correlates of criminal behaviour (Gendreau, Little & Goggin, 1996). Although the information available for study was limited, we were able to determine that individuals denied a pardon evidenced many known risk factors. They were more likely to be male, offended at an earlier age, had more prior convictions and were more likely to have been incarcerated. We cannot say for certain that ex-offenders granted a pardon were less likely to reoffend. However, our findings on the limited risk factors available for study all point in the direction of higher risk for the denied applicants.

Demographic information revealed that most of the pardon applications were received from a handful of provinces and that differences existed in the number of pardons granted and denied in British Columbia and Quebec. Perhaps these differences were the result of different levels of public awareness regarding the pardon criteria and process from province to province. Also, the time it required to grant a pardon was significantly less than the time to deny a pardon. This difference was consistent across the provinces. This is most likely due to the fact that individuals whose applications do not appear to meet the criteria to grant a pardon are so informed and allowed to provide representation, either oral or written, to the NPB before a final decision is made. Given that the criminal records of the granted sample contained fewer factors that are known to be associated with criminal behaviour than the denied sample it appears that the criteria for granting pardons is consistent with research regarding an offender's likelihood to reoffend. To further examine the decisions to grant pardons the second study looked at the long-term outcomes of the granted sample.

Study 2: The Outcome of Pardons

The purpose of Study 2 was to examine the outcomes of granted pardons. A longitudinal design would have been the most appropriate methodology to study long-term outcomes. However, one could not be conducted mainly because of the manner in which the files were stored. Successfully granted files were kept separate from those that were revoked making it impossible to randomly select a representative sample of pardon files that would include both pardons that were successfully granted as well as those that had been revoked. Therefore, two independent samples were constructed. From Study 1, 603 of the 1,128 offenders showed no evidence of further criminal activity between the years 1989 and 1996. The remaining 525 offenders were granted pardons and subsequently had their pardons revoked (55.2%) or ceased to be in effect (44.8%) during the same time period. This second group represented an over sampling of revoked cases. Since revocations represent 2.6% (6,046/234,779) of all pardons granted or issued, it was necessary to over sample the revocations in order to create a satisfactory sample size for statistical analysis.


There were few personal-demographic differences between the revoked and the non-revoked samples. Both groups were similar in age at first conviction and when applying for a pardon. Also, there were no significant differences in the number of times they had been previously denied. In fact only two differences were found. First, the offenders whose pardons were revoked were more likely to be unemployed (70.7% vs. 55.6%, X = 27.3, p < .001) and second, the revoked group had fewer females (10.5% vs. 19.6%, X = 17.9, p < .001).

Table 3 compares the revoked and non-revoked samples along a few criminal history variables. The criminal histories of the revocation sample were more extensive than the non-revoked sample. For example, the revoked offenders had more prior convictions which included more sexual convictions and more non-sexually violent convictions compared to the non-revoked sample. In contrast, when examining their most serious offence prior to receiving a pardon no differences were found between the two groups relating to violent crimes (against person or sex offences). Reviewing the offences precipitating the revocations we found that the revocations were most often the result of liquor and traffic violations (35.8%). In addition, 20.4% were revoked as a result of a property offence and 18.9% following a crime against person (excluding sex offences). The average length of time between receiving a pardon and its revocation was 4.2 years (SD = 1.7 years).

Table 3. Criminal Histories of Revoked and Non-Revoked Pardons
Variable Non-Revoked Revoked t/X p
  (n = 603) (n = 525)    
# of prior convictions (mean) 1.8 2.8 7.9 ***
# of prior offences (mean) 2.6 3.9 4.1 ***
Any prior sexual offences (%) 2.7 5.1 4.7 *
Any prior non-sexually violent offences (%) 12.9 19.6 9.3 **
Most Serious Prior Offence (%)
Property 45.4 38.1 6.2 *
Liquor/Traffic 21.9 27.8 5.3 *
Narcotics 16.6 13.9 1.6 n.s.
Person 7.8 10.5 2.5 n.s.
Sexual 2.0 3.0 1.3 n.s.
Other 6.3 6.7 0.1 n.s.

Note: *** = p < .001, ** = p < .01, * = p < .05, n.s. = nonsignificant


Few differences in the personal-demographic variables were observed. The revoked sample had more males and was more likely to be unemployed than the non-revoked sample. Both of these factors are correlates of recidivism (Gendreau et al., 1996). Some criminal history risk factors were also found associated with the revoked offenders. Similar to the findings of the first study, Study 2 found that the unsuccessful group (revoked) had committed significantly more offences resulting in more convictions than the successful group (non-revoked). Individuals whose pardons were revoked were more likely to have histories of violent offences (sexual and non-sexual) although no differences were found among crimes against person for the most serious prior offence. Notable was the fact that liquor and traffic violations along with property offences accounted for more than half (56.2%) of the post-pardon convictions resulting in revocations.

Although there were no differences found in sex offending in terms of their most serious prior offence, differences in the number of prior sex offences were found between the non-revoked and revoked samples. Given that sex offenders are often cause for concern due to the nature of the offence and the seriousness of the harm caused to victims, Study 3 explored in more detail the sexual offending among the revoked offenders.

Study 3: Pardoned sex offenders

The protection of children and other vulnerable groups from sexual victimization is an important public concern. Researchers have responded by studying the factors associated with the causes and treatment of sexual offending. Previous research has found that sex offenders' general recidivism rates are comparatively low but that sex offenders are more likely to reoffend sexually than non-sex offenders (Hanson, Scott & Steffy, 1995). Since a pardon results in the sealing of an offender's criminal record it becomes difficult to determine whether an individual had been previously convicted of a sex offence. This may be especially important information for agencies hiring individuals to work with children and/or vulnerable adults. When screening candidates for employment or volunteer positions, an individual who has previously been convicted of a sex offence and granted a pardon would not be detected under the current screening system. Therefore, it is important that the criteria for granting pardons effectively differentiate between offenders (sex offenders and non-sex offenders) who have demonstrated sustained law-abiding behaviour resulting in successful reintegration and those who have not.

The exact number of pardoned sex offenders is not known but estimates can be derived from existing offender databases. The RCMP has compiled over 2 million Finger Print Service records between April 1, 1970 and March 31, 1999. The percentage of recorded charges for sexual offences found within these files is 5.5 (130,385). The RCMP also maintains a separate database of adult offenders pardoned since 1992. This database contains over ninety thousand active files of which 4,456 (4.9%) contain sex offence charges (RCMP, personal communication, September 1999). It is important to note that these numbers (5.5% of the general offence population and 4.9% of the pardoned records since 1992) represent not only sexual offence convictions but also charges relating to sex offences that were not proceeded with (e.g., stayed or withdrawn). In order to provide a more accurate estimated total of pardoned sex offenders and their rates of sexual recidivism, Study 3 analyzed the sexual offending, pre and post pardon, of revoked offenders (n = 525). In the measurement of recidivism, adult (18 years or older) prostitution offences were excluded from our definition of sexual recidivism (in total, only 4 cases were excluded).


Table 4 illustrates the incidence of sexual offending among the revocation sample. As found in Study 2, 5.1% (27/525) of the revoked sample had been convicted of a sex offence prior to being granted a pardon. Further, 32 of 525 (6.1%) revocations were for sexual recidivism. The rate of post pardon sex convictions among the non-sex offenders was 4.4% (22/498); among the sex offenders the rate was 37% (10/27; c = 47.61, df = 1, p > .001). However, it is important to note that 63% (17/27) of the revoked sex offenders reoffended in a manner other than sexual. The percentages of sex offenders found in our revocation sample, 5.1% pre-pardon sex offenders and 6.1% post-pardon sex offenders, are consistent with the percentage of sex offenders found in the RCMP databases, strengthening the generalizability of our sample.

Table 4. Sex offending of revoked offenders (n)
  Post Pardon Sex Offence
Prior Sex Offence Yes No Total
Yes 10 17 27
No 22 476 498
Total 32 493 525

Based on these findings, estimates can be made as to the prevalence of sexual offending among pardoned offenders. Over a 28-year period (1970-1998) 234,779 pardons were granted or issued of which 2.6% (6,046) have been subsequently revoked. Therefore, 228, 733 (97.4%) pardons were successful (i.e., not revoked). From Study 2, the non-revoked sample contained very few sex offenders (2.7%; 16/603). Narrowing the definition to exclude adult prostitution offences, the number of sex offenders who did not have their pardons revoked was reduced from 16 to 12 yielding an estimate of 2.0%. Assuming that the current sample is representative of the general pardoned population, the estimated number of sex offenders granted a pardon and who were not revoked since 1970 would be approximately 4,575 (2% of 228,733) as illustrated in Table 5.

Using data from the revocation sample, 5.1% of the offenders were convicted of a sex offence prior to receiving a pardon. When this percentage is applied to the total number of actual revocations during the 28-year period (5.1 % of 6,046), we project that there were approximately 308 revocations for offenders pardoned for a sex offence. Therefore, combining the estimated number of sex offenders granted a pardon but not revoked (4,575 or 2.0%) with the estimated number of revoked pardons containing previous sex convictions (308 or 5.1%) we arrive at a total number of 4,883. It is estimated that 2.1 % of all pardons were granted to sex offenders (4,883/234,779) over a 28-year period.

Table 5. Estimated prevalence of sex offenders granted a pardon (n)Footnote 3
  Previous Sex Offence
Successfully pardoned Yes No Total
Yes (non-revoked) 4,575
No (revoked) 308
All pardons 4,883

A closer examination of the sex offenders in the revocation sample indicated that 10 of the 27 reoffended sexually. This would suggest that slightly over one third (114/308) of the estimated revocations for previously pardoned sex offenders would recidivate sexually. Recall that it was projected that 4,883 sex offenders were granted a pardon. A summary of the population estimates is illustrated in Figure 3.

Figure 3. Estimates of Sex Offenders in the Pardoned Population
  Actual # of pardons # of estimated sex offenders
Granted/Issued 228,733 4,575
Revocation/Cessation 6,046 308
Total Pardoned 234,779 4,883 (2.1%)

The resulting overall estimate of sex offenders granted a pardon who are likely to be revoked for a new sex offence is approximately 2.3% (114/4,883).


Public safety concerns often focus on sex offenders and their potential to cause serious harm. As a result, the pardon process has been called into question as it relates to sex offenders. However, the extent to which sex offenders receive pardons and reoffend sexually has been largely unknown until now. Study 3 examined the sex offending of a sample of pardoned offenders in order to provide population estimates. The results indicated that approximately 2.1% (4,883) of all pardons (234,779) would have been granted to sex offenders. Furthermore, 2.3% of the projected number of pardoned sex offenders, that is 114, were estimated to reoffend sexually. Assuming that an equal number of pardoned sex offenders were reconvicted of sex crimes each year from 1970 to 1998 this would amount to four sexual recidivists per year. Compared to an annual average of approximately 31,500 (range 35,524 - 28,952) sexual incidents reported to the police between 1994 and 1998 (Statistics Canada, 1999) four cases appears small. Also, during the same time frame (1994/95-1997/98), the average number of adult sexual convictions reported by Statistics Canada was about 2,800 (range 2,725 - 2,881; data extracted from the Adult Criminal Court Survey, Statistics Canada, 1997, 1998). Taking these figures into account and considering that over a quarter of a million offenders have been granted a pardon between 1970 and 1998 the number of sex offenders pardoned and likely to be convicted of another sexual offence is comparatively small. Nevertheless, additional strategies to reduce the risk of sexual victimization can be sought.

Research has shown that when known sex offenders reoffend they have a higher probability to reoffend sexually than non-sex offenders. While non-sex offenders have a higher recidivism rate overall they are almost twice as likely to commit non-sexual violent crimes than sex offenders (Hanson et al., 1995). Refusing pardons to sex offenders would not completely eliminate the possibility of a pardoned offender committing a sex offence resulting in a revocation. In Study 3, 4.4% (22/498) of non-sex offenders granted a pardon were later convicted of a sex offence and consequently had their pardon revoked. Therefore, since the majority of sex offenders are not reconvicted of a sex offence it would run counter to the original intent of a pardon (the forgiveness for past wrongdoings), to deny pardons to all individuals who have been previously convicted of a sex offence. However, given the potential for harm posed by sex offenders, invoking special measures may be warranted. For example, flagging sex offenders' criminal records for volunteer screening purposes, as recommended by the High Risk Offender F/T/P Working Group (Solicitor General Canada, 1998), to increase public safety.

Summary and Conclusions

We learned from Study 1 that the criteria used to make pardon decisions are consistent with some of the elements associated with continued criminal conduct. Offenders with less serious criminal histories were more likely to be granted a pardon while those denied a pardon had more previous convictions of a sexual and/or violent nature. The factors differentiating offenders granted and denied pardons were also consistent with the research on the risk factors relating to future recidivism (Gendreau et al., 1996). From the available evidence we are led to conclude from Study 1 that pardon decisions are made on empirically defensible grounds.

An examination of the outcomes for revoked and non-revoked cases in Study 2 revealed few differences. Only two personal demographic factors showed significant differences, gender and employment. However, among the criminal history indicators, differences were found in the number of prior convictions and prior sexual and violent offences. Although few differences were found, one must bear in mind the limited data available for analysis, the results, nevertheless, indicated that the criteria to grant pardons are consistent with known risk factors as would be expected.

Finally, Study 3 found that sexual offending accounted for approximately 5% of the pardoned offences and approximately 6% of the recidivism offences for the revoked sample. The observed frequency of sex offending in our sample was applied to the entire pardoned population to derive expected sexual offending rates. Based on these sample, it is estimated that among 234,779 pardons granted, 4,883 were granted to sex offenders with 114 reoffending sexually. Notwithstanding the small percentage that this represents the possibility of reoffence remains a concern. Hence, special measures for pardoned sex offenders may be a more effective alternative than denying pardons to all offenders previously convicted of a sex offence. One alternative to the blanket denial of pardons to sex offenders that had been proposed is to flag a sex offender's pardoned criminal record so that under specified circumstances this information may be made available (Solicitor General of Canada, 1998).

Although the studies presented contained reasonable sample sizes that were randomly chosen, a cautionary note is in order. The over sampling of revocations may cause the prevalence of sex offending to be over estimated. Future studies containing revocation samples proportionate to the samples of granted pardons may produce slightly different findings. The personal demographic and criminal history information available for our studies was limited. More detailed information related to factors empirically found associated with an offender's risk to recidivate would enable a more comprehensive evaluation of pardon decisions. Also, in order to facilitate research examining the recidivism of pardoned sex offenders a more detailed account of personal demographic information of the victims and information relating sexual offending would be desirable.Footnote 4


Burton, V. S., Jr., Travis, L. F., & Cullen, F. T. (1988). Reducing the legal consequences of a felony conviction: A national survey of state statutes. International Journal of Comparative and Applied Criminal Justice, 12(1), 101-109.

Canadian Centre for Justice Statistics(1997). Adult criminal court statistics, 1995-96 (Tech. Rep. No. 85-002-XPE, Vol. 17, no. 14). Ottawa, ON: Statistics Canada.

Canadian Centre for Justice Statistics(1998). Adult criminal court statistics, 1996-97 (Tech. Rep. No. 85-002-XIE, Vol. 18, no. 7). Ottawa, ON: Statistics Canada.

Canadian Centre for Justice Statistics(1998). Adult criminal court statistics, 1997-98 (Tech. Rep. No. 85-002-XPE, Vol. 18, no. 14). Ottawa, ON: Statistics Canada.

Canadian Centre for Justice Statistics (1999). Crime statistics in Canada, 1998 (Tech. Rep. No. 85-002-XIE, Vol. 19, no. 3). Ottawa, ON: Statistics Canada.

Canadian Centre for Justice Statistics (1999). Sex offenders (Tech. Rep. No. 85-002-XIE, Vol.19, no. 3). Ottawa, ON: Statistics Canada.

Couillard, D. (1985). The royal prerogative of mercy in Canada: Its origins, its evolution, its issues. Ottawa: Secretariat of the Solicitor General, Solicitor General Canada.

Criminal Records Act, R.S.C. 1985, Chap. C-47.

Criminal Records Act, R.S.C. 1985, Chap. C-47, amended 1992, c.22, ss.1-10.

Gendreau, P., Little, T., & Goggin, C. (1996). A meta-analysis of the predictors of adult offender recidivism: What works! Criminology, 34(4), 575-595.

Hanson. R. K., Scott, H., & Steffy, R. A. (1995). A comparison of child molesters and non-sexual criminals: Risk predictors and long-term recidivism. Journal of Research in Crime and Delinquency, 32(3), 325-337.

National Parole Board. (1998). Royal prerogative of mercy. Ottawa, ON: Author.

National Parole Board. (1999). Performance monitoring report 1998-1999. Ottawa, ON: Performance Measurement Division.

Smith, A. T. H. (1983, Autumn). The prerogative of mercy, the power of pardon and criminal justice. Public Law, (389-439). London: Stevens & sons.

Solicitor General Canada. (1981). Clemency review: Issues paper. Ottawa, ON: Policy Branch.

Solicitor General Canada. (1997). Consultation document on proposals to amend the Criminal Records Act. Ottawa, ON: Corrections Directorate.

Solicitor General Canada. (1998). Information systems on sex offenders against children and other vulnerable groups: Report to Federal, Provincial and Territorial Ministers. Ottawa, ON: The F/P/T Working Group on High Risk Offenders.

Street, L. (1996). The screening handbook: Protecting clients, staff and the community. Ottawa, ON: Canadian Association of Volunteer Bureaux and Centres.


  1. 1

    Bill C-7 was passed by Parliament on February 14, 2000. It amends the CRA to provide a legislative framework allowing access to the pardoned records of certain sex offenders.

  2. 2

    Study 2 includes a sample that were either revoked (pre/post CRA amendments) or ceased to be in effect (post CRA amendments only) that will hereafter be referred to as revoked or revocations.

  3. 3

    These extrapolations are lower than previously presented in a report prepared by the Federal/Provincial/Territorial Working Group on High Risk Offenders entitled "Information Systems on Sex Offenders Against Children and Other Vulnerable Groups" which was based on a pilot study (Solicitor General Canada, 1998). The current findings and resulting population estimates are based on larger samples and an improved research design. Therefore, the present results are considered to be more representative of the pardoned population.

  4. 4

    NPB is developing an automated pardon information system (Pardon Application Decision System - PADS) that will systematically provide offender data when it is implemented in July, 2000.

Date modified: