Fifty-four (11 percent) of the youth reported experiencing a nonfatal drug overdose in the previous 6 months. Multivariate analysis indicated that being female, using crystal methamphetamine (noninjecting), injecting heroin, and injecting cocaine were linked with nonfatal overdose events among study participants. Just over half of the cohort reported being aware of the potency of the drugs consumed prior to their most recent overdose. The authors suggest that the high rate of drug overdose can be reduced by modifying the risk environment of street-involved youth through such interventions as supervised drug injection sites and the provision of supportive housing. Street-based outreach efforts were conducted in recruiting street-involved youth into the study, called the At-Risk Youth Study, which was conducted between September 2005 and June 2006. Eligibility for participation was restricted to youth ages 14 to 26 who reported using drugs other than marijuana in the last 30 days. The following sociodemographic, behavioral, and drug-use variables were examined as potential predictors of nonfatal overdoses: age, gender, ethnicity (White and "other"), requiring help injecting, injecting in public, noninjection crystal methamphetamine use, injection heroin use, injection cocaine use, injection crystal methamphetamine use, noninjection heroin use, noninjection opiate use, and frequent liquor use. Participants who reported experiencing an overdose were also asked to identify the main drug consumed prior to the overdose. Pearson's chi-square test and Fisher's exact test were used to determine factors associated with experiencing an overdose event in the previous 6 months in unadjusted analyses. In addition, a fixed multivariate logistic regression model was used, which included all of the variables considered in univariate analysis. 2 tables and 11 references