Among studies cited by several medical associations as a guide for warning parents about the negative effects of television was Brandon Centerwall’s (1992) analysis of the effect of the invention and distribution of television on homicide rates. Centerwall claimed that the introduction of television substantially increased homicide rates in the United States and Canada, remaining stable in South Africa until the ban on television was lifted. This study examined other alternative explanations for increases and decreases in homicide rates. This study tested an alternative hypothesis using a multivariate time-series analysis that examined relationships involving primary groups to understand variations in homicide over time. This was not to rule out media effects but to compare the variation in social learning processes within intimate personal groups. The analysis consistently challenged the view that television was responsible for a significant proportion of increases in homicide following its invention. In the United States, family breakdown and alcohol use were crucial correlates of homicide rates. In a multivariate analysis, a positive correlation between homicide rates and the spread of television disappears. For Canada and South Africa, the results were similar. The findings constitute a serious challenge to Centerwall’s thesis but continue to support traditional sociological perspectives. Tables and references