To evaluate the validity of 2 self-report methods for estimating cocaine use, Timeline Follow-Back (TLFB) and weekly calendar reports from 65 patients with a cocaine use disorder were compared with urine drug test results. The TLFB showed fair to moderate validity, and the weekly calendar showed moderate to high validity in measuring the frequency of cocaine use. Similar results were obtained when the self-report measures were used to time specific cocaine use episodes. In addition to evidence for superiority of the weekly calendar, the validity of self-reports was inversely related to the percentage of positive urine test results. Furthermore, there was some evidence that validity increased as the time window over which the comparisons were drawn increased. Given the central role of self-reports in the clinical and research evaluation of drug use, factors affecting their validity warrant further investigation.