Drug use in the Navy and the civilian sector, the source of new Navy personnel, continues to concern Navy managers. The Navy currently conducts a random urinalysis program to deter and detect drug abuse. Navy field commanders, who are responsible for execution of the testing program at their units, have latitude in both the frequency and rate of drug testing. The entire program is governed by strict procedural and legal safeguards. However, no systematic analysis of the sampling program has been undertaken to discover cheaper, more effective alternatives. This paper develops probabilistic models for urinalysis testing. Several observations are made concerning current practice, both with and without gaming. An alternative testing procedure is analyzed. This alternative procedure is the basis for a new microcomputer-based random urinalysis strategy currently under development by the U.S. Navy.