We present a prospective, controlled study of the association between preemployment drug screening results and employment outcomes in 2537 postal employees. For identified marijuana users, relative risk for turnover was 1.56 (95% confidence interval [Cl], 1.17 to 2.08); accidents, 1.55 (95% Cl, 1.16 to 2.08); injuries, 1.85 (95% Cl, 1.30 to 2.64); and discipline, 1.55 (95% Cl, 1.03 to 2.32). Their mean absence rate was 7.1% compared with 4.0% for nonusers. For identified cocaine users, relative risk for turnover was 1.15 (95% Cl, 0.65 to 2.05); accidents, 1.59 (95% Cl, 0.95 to 2.67); injuries, 1.85 (95% Cl, 1.01 to 3.39); and discipline, 1.40 (95% Cl, 0.62 to 3.17). Their mean absence rate was 9.8%. Our study shows that a preemployment drug screen positive for marijuana or cocaine is associated with adverse employment outcomes. The level of risk, however, is much less than previously estimated. This finding has important implications for the social, legal, and economic arguments for and against drug testing.