How frequently should urine samples be collected and analyzed to accurately measure drug use in clinical trials of cocaine abuse treatments? Previous research suggests that analyzing one of three weekly urine toxicology samples in an opiate-related trial may be sufficient. To empirically address this question in the field of cocaine research, we examined the weekly variation in the cocaine metabolite benzoylecgonine (BE) concentration between pairs of weekly urine samples from a clinical trial of a treatment for cocaine dependence. Twice weekly urine samples from 71 subjects collected over eleven weeks were assayed for quantitative BE levels. Agreement between pairs of samples was estimated for both quantitative and qualitative measures of BE. Results indicated substantial intra-week variation with correlations never exceeding .50 and approximately 20% disagreement among samples using cutoff values in place of quantitative levels. Both samples, however, supported similar conclusions about group-level behavior.