The review focused on three issues: (1) underreporting of drug use on direct questioning; (2) omission of groups who are at risk; and (3) procedures for estimating the use of drugs such as heroin, which have low rates of use. The papers discuss validity as a multidimensional concept, research conditions and respondent characteristics affecting levels of underreporting, advantages and limitations of school surveys, and results of urine testing to validate self-reported use of marijuana by pregnant women. Other papers present a historical outline of the different techniques used to derive national estimates of heroin users; the nominative technique, in which respondents are asked about drug use in a friend; dynamic simulation models; and the multiple recapture census. The papers present results of studies using these techniques in the United States and Canada. Included are results from NIDA's National Household Survey of Drug Abuse. Figures, tables, chapter reference lists, and an address list of the participants in the technical review.