This proposal addresses a growing body of concern about the validity of survey responses to questions about drug use. It is also a response to the perceived need for more information about drug use in special populations that may be particularly at risk for this behavior. Three years of funding are sought to support follow-up secondary analyses on two existing data sets, the National Longitudinal Survey of Youth (NLSY) and the Juvenile Drug Use Forecasting Program (JDUF). Analyses of the NLSY data will follow-up and extend previous research on methodological issues in substance use reporting by evaluating correlates of diminished reporting over time and correlates of inconsistencies with respect to the timing of first drug use. JDUF analyses will examine the validity of self-reports and the correlates of drug use underreporting. Analyses will examine period, cohort and other subject influences on reported and detected drug use across multiple interview sites over a six year period. Analyses will also evaluate the dimensionality of lifetime drug use reports, the role of alcohol and marijuana as gateway substances and correlates of drug use levels. This project will employ innovative analytic strategies for handling longitudinal and clustered cross sectional data. Monographs generated from both sets of analyses will inform the future design and analysis of epidemiological research on drug abuse. Findings will be discussed in terms of their relevance for cognitive theories of reporting behavior. JDUF analyses will be used to draw conclusions about the future direction of prevention and treatment planning for high risk adolescents.