We evaluated inconsistencies in responses to questions about lifetime cocaine and marijuana use asked of nearly 10,000 respondents from the United States in the National Longitudinal Survey of Youth in 1984 and 1988. Our analyses showed that 14% of all responses on cocaine use and 17% of all responses on marijuana use were inconsistent in some way. The types of inconsistencies varied according to the substance; cocaine reports yielded more inconsistencies with regard to timing of first use, while for marijuana most of the inconsistencies were with respect to use disclosure. For both substances, lower level users were more likely to be inconsistent in their reports of drug use. Alternative methods for handling inconsistencies affected estimates of incidence and prevalence. Inconsistencies also varied according to respondent race/ethnicity. Implications of these findings for program evaluation are discussed.