Issues addressed are the rationale for drug testing and the choice of tests, techniques and difficulties of implementing drug testing in various field situations, interpretation of drug test results, and the contributions of drug testing vs. self-reports to arriving at study conclusions. The studies include experiences with urinalysis, hair analysis, and saliva analysis. Four groups of subjects were interviewed and biological specimens were obtained. Study 1 subjects provided urine specimens. Study 2 subjects provided urine and hair specimens. Study 3 subjects provided a hair specimen. Study 4 subjects provided hair and saliva specimens. Study 1 results showed that urinalysis appeared to have only limited utility for validating or supplementing self-reports of drug use in field studies. Study 2 indicated that hair analysis was significantly but not greatly superior to self-reports and urinalysis in identifying drug ingestion during a prior 30-day period for a sample of addicts in treatment. As a whole, Study 2 bolstered confidence in the utility of hair analysis for validly identifying hidden drug use in field studies. Hair testing in Study 3 altered the substantive conclusions that would have been drawn from self-reports alone, both in terms of estimating cocaine/crack use prevalence and documenting the associations between cocaine/crack use and social factors of importance. Study 4 results showed that hair analysis provided a reasonable estimate of the true prevalence of cocaine/crack use, but the window of detection for saliva analysis was too narrow to provide an estimate of cocaine/crack use for the previous month. Biological testing for drugs of abuse can make a significant contribution to improve behavioral drug-abuse research by improving validity of studies that would otherwise rely entirely on self-reports. Hair testing is applicable when the period of interest is drug use within the last 1 to 3 months. Saliva is applicable for determining use within the last few days, especially when urine collection is not feasible or is inconvenient. 4 tables, 39 references