Outlines 5 models of the temperature-aggression hypothesis: negative affect escape, simple negative affect, excitation transfer/misattribution, cognitive neoassociation, and physiological-thermoregulatory. Reviews relevant studies. Aggression measures include violent crime, spouse abuse, horn-honking, and delivery of electric shock. Analysis levels include geographic regional, seasonal, monthly, and daily variations in aggression, and concomitant temperature-aggression effects in field and laboratory settings. Field studies clearly show that heat increases aggression. Laboratory studies show inconsistencies, possibly because of several artifacts. Specific models have not been adequately tested, but the excitation transfer/misattribution and cognitive neoassociation approaches appear most promising, whereas the negative affect escape appears the least viable. Suggestions for future work are made.