Surveys of drug use are continually criticized on the premise that respondents underreport the extent of their drug use. Validation studies conducted prior to the mid-1980's that have involved known samples of drug users or urinalysis techniques showed that drug use was fairly accurately reported in self-report surveys; however, more recent validation studies conducted with criminal justice clients, using improved urinalysis techniques, suggest less concordance between urinalysis and self-report. Based on this review of relevant studies the author concludes that valid self-reporting of drug use is a function of the recency of the event, the social desirability of the drug, and nuances of data-collection methodology. Some methods used to improve the accuracy of self- reporting in drug-use surveys are assurances of anonymity and confidentiality and the use of self-administered questionnaires that involve no contact with an interviewer and no means of identifying which respondents provided particular information received. Overall, there are indications of a small underreporting bias in the National Household Survey on Drug Abuse and the Monitoring the Future Survey, but its overall effects are relatively small. 3 figures and 33 references