The sample was recruited during 1994 by means of advertisements placed in methadone units, drug-free residential treatment agencies, needle exchanges, rock magazines, local papers, and by word of mouth. The mean age of the sample was 30.2 years. In order to obtain heroin users at various stages in their careers, purposive sampling was used to obtain approximately equal representations of users in and out of treatment for opiate dependence. The structured interview devised examined demographics, drug-use history, heroin dependence, personal experience with overdose, presence at other peoples' overdoses, impediments to help-seeking, prevention of overdose, beliefs and attitudes regarding overdose, and attitudes toward naloxone. For continuous variables, t-tests were used. Categorical variables were analyzed using x2, and corresponding odds ratios and 95 percent confidence intervals were calculated. Findings show that the experience of overdose was widespread, with two-thirds of subjects (68 percent) reporting having overdosed. The median number of lifetime overdoses was three, with males and females equally likely to have overdosed. The majority (62 percent) of most recent heroin overdoses occurred in conjunction with the consumption of other central nervous system depressants (alcohol, benzodiazepines and other opioids). Logistic regression analyses indicated three independent factors associated with having overdosed: longer heroin-using careers, greater heroin dependence, and higher levels of alcohol consumption. Implications for the reduction in the prevalence and frequency of overdose are discussed. 4 tables and 30 references