Although several researchers have examined the validity of self report
ed drug use, few studies have focused on those factors that affect val
idity. The present investigation uses Drug Use Forecasting data to ide
ntify predictors of valid self-reporter among arrestees from five citi
es in the United States. The multivariate results show that factors af
fecting the misrepresentation of drug use vary by site and by drug cat
egory. Explanations for these findings are offered and suggestions for
improving the validity of arrestees' self-reports are discussed.