Daily records were randomly generated for a period of 128 patient years (totaling 46,752 separate days) in order to provide a sufficiently large sample from which to examine the relationship between urine testing patterns and drug-use detection. A random number generator was used to determine drug use on a daily basis, assuming six frequencies of drug use. A separate randomization was performed for each drug-use pattern. A random number generator was also used to determine those weekdays (Monday through Friday) on which urine tests would be performed, assuming five drug-testing patterns. For a given drug-testing pattern, urinalysis results were determined for each of the six drug-use patterns, assuming three drug-detection windows (24 hours, 48 hours, and 72 hours). The findings indicated that when urine testing was performed at a rate of eight times per year, the probability of testing positive in a given month was little better than 50-50, even for daily use. Infrequent drug use was difficult to detect regardless of drug-testing frequency; and the benefits of more frequent drug testing were greatest with moderate drug use. These data provide a guide for clinicians to use in matching drug screen schedules to frequency or pattern of suspected drug use. 3 figures, 2 tables, and 4 references