Harries and Stadler (1988) observed only a positive linear trend, with no curvilinear relationship, between heat and violence in their 1980–81 Dallas field data. When Cotton (1986) observed similar trends in his data, we noted that in fact the variance in violence increases with temperature (Bell & Fusco, 1986). We report here similar statistically significant heteroscedasticity in the Harries and Stadler data. We propose that both aggressive and escape tendencies increase with high ambient temperatures, such that in some circumstances aggression will decline under very hot and uncomfortable conditions. We conclude that the negative affect escape model of the temperature-aggression relationship is neither dead cold nor hot and bothered.