This formula was applied to choose a number of sample units for routine chemical analysis of controlled substances, with the analytical result considered to be representative of the entire lot of powdered drug units in the exhibit. During this study, 85 criminal cases, containing 173 drug exhibits with a total of 19,994 powdered units, generated 238 different populations; over 95,000 analytical tests were conducted. The formula found to be most appropriate for selecting a representative sample of powdered street drug exhibits was n = 20 + 10% (N-20), where n equals sample size and N equals total population (N is greater than or equal to 20). This sampling leads to forensic analyses of controlled substances which are supported to a reasonable scientific certainty and are obtained efficiently. 3 figures, 4 tables, and 5 references