Ninety cases of probationers were monitored for drug use for 6 consecutive months. Hair and urine specimens were obtained at approximately 1 month intervals and assayed for several psychoactive drugs, including cocaine. Self-reported, long-term drug use was also surveyed among these persons. The Bayes Theorem was employed to examine the data for conditional probabilistic outcome patterns. The data reported in this study supported the accuracy of cocaine assays based upon hair analysis. Based on their self-reported use, this population had evidence to support a generalization of being at moderate to high risk for drug use and drug exposure. The concordance outcome patterns gave strong support to the overall effectiveness of the hair assay for cocaine. This support was demonstrated in two ways: (1) the failure to uncover any meaningful incidence of persons testing positive in urine but negative in hair; and (2) there was a correspondence between the likelihood to test urine positive for multiple longitudinal urine tests, and the likelihood to be hair positive for multiple samples. When considering cocaine, the greater the number of sequential positive urine tests, the more likely that the hair assays will have higher rather than lower assay values. When those few cases that show a nonconcordant outcome for hair and urine assays for cocaine are examined, it is found that they do not represent absolute conundrums, but rather have either straightforward clinical explanations or else they have at least plausible explanations, such as cocaine appearing in the hair, but after a longer than expected lag between ingestion and appearance above the scalp. Hair analysis for cocaine is a useful and effective method for estimations of prevalence in populations. There does not yet appear to be any insurmountable obstacles to its employment in that capacity. 16 tables, 25 references