Disagreement over the mechanism of death in so-called heroin overdose has been noted in the medical literature for more than a decade. We studied the relationship between the potency of street heroin and the frequency of heroin overdose deaths in San Antonio in order to test the“pharmacologic overdose”hypothesis. We found a small, nonsignificant correlation (r= +.13) between monthly mean heroin dose and monthly number of heroin overdose deaths over a 5-year period. This low correlation contradicts findings of previous studies in Washington, D.C. and Atlanta, conducted over shorter periods of time, in which high positive correlations were found. The low correlation does not support the pharmacologic overdose hypothesis. Since there were confounding variables, the data do not refute the hypothesis. These variables include the use of alcohol and other depressants, hidden suicide, and reduced opioid drug tolerance due to institutional treatment.