Members of the experimental group were told that their specimens had a one-in-three chance of being tested; all subjects received their test results. The results showed that testing a one-third random sample of collected specimens did not lead to increased drug use or to a reduction in self-disclosure validity over a one-month period. These effects were not changed when other potential influences were examined, including race, age, probation officer, or offense. The findings called into question the conventional assumption that testing and feedback deter drug use. The study suggests that it may be possible to redesign drug-testing programs to decrease cost while maintaining effectiveness.