Thermal stress was measured in terms of both ambient temperature and a discomfort index that takes into account the influence of humidity acting in concert with temperature. Regression analyses were performed in two stages. In the first, data for all neighborhoods and all days of the study period were combined into ambient and discomfort index models. In the second, models differentiated among three levels of neighborhood socioeconomic status. With weekends controlled, both the discomfort index and ambient temperature were significant independent variables overall and in medium and low status neighborhoods. In addition to confirming the Anderson and Anderson (1984) finding that a weekend effect was the most powerful predictor of assault frequency, results also indicate that the relationship between thermal stress and aggression is linear, rather than curvilinear, and that a reduction of aggression with increasing temperature does not occur within the normal range of temperatures. 3 tables and 11 references. (Author abstract modified)