Nicholaichuk et al. have proposed a new method of constructing post-hoc comparison groups in which cases are selected from existing criminal history records. The method is intuitively plausible, relatively simple, and potentially applicable to diverse treatment settings, at least in Canada where the study was conducted. The researchers used recidivism information drawn from the national criminal history records maintained by the Royal Canadian Mounted Police (RCMP). Although the RCMP records are the most comprehensive records available in Canada and are used for official court purposes, they do not include all offenses nor all offenders. Records are routinely purged due to death, official pardons, and extended periods of inactivity. The obvious effect of the purging policy is that the older records should include a disproportionate number of recidivists. This biasing effect should be negligible in the first 10 years, but would be expected to increase with longer follow-up periods. This artifact in RCMP records could introduce a bias in Nicholaichuk et al.'s outcome study, because the treatment and comparison groups were linked to the recidivism information by using different procedures. The subjects in the treatment groups were identified from clinical records of the Clearwater Program. If they had no subsequent convictions on their RCMP records, they were considered non-recidivists. The subjects in the comparison group, however, were selected from existing (active) RCMP records, which would be expected to include a disproportionate number of persistent offenders. 1 figure and 12 references