BACKGROUND:Research has shown a strong link between alcohol use and a variety of problems, including violence. Parker and colleagues have presented a selective disinhibition theory for the link between alcohol use and homicide (and other violence) that posits a causal relationship that is also influenced by other situational and contextual factors. This model is particularly well suited for aggregate-level investigations. In this study, we examine the impact of alcohol factors, including consumption measures and Alcoholics Anonymous (AA) membership rates, on homicide mortality rates in Ontario, and test predictions derived from the selective disinhibition model. METHODS:Time series analyses with ARIMA modeling were applied to total, male, and female homicide rates in Ontario between 1968 and 1991. The analyses performed included total alcohol consumption, spirits consumption, beer consumption, and wine consumption. Missing AA membership data were interpolated with cubic splines. RESULTS:For the total population and males, homicide rates were significantly and positively related to total alcohol consumption and to the consumption of beer and spirits. They were also negatively related to AA membership rates in the analyses involving spirits and wine and positively related to unemployment rates in the analyses involving beer, wine, and total alcohol. Among females, none of the measures were significant predictors of homicide mortality rates. CONCLUSIONS:These data provide important support for the selective disinhibition model and confirm important relationships between per capita consumption measures and homicide mortality rates, especially among males, seen in other studies. Additionally, the results for AA membership rates are consistent with the hypothesis that AA membership and treatment for misuse of alcohol can exert beneficial effects observable at the population level.