Sixty undergraduate males delivered electric shocks to a confederate as punishment for making errors on a learning task. Half did so after being exposed to an aggressive model, whereas the other half did so without being exposed. Shocks were delivered from a room whose atmosphere was either unpolluted or contaminated by a moderately unpleasant odor (ethyl mercoptan) or an extremely obnoxious stench (ammonium sulfide). As hypothesized, the moderately unpleasant pollutant facilitated higher levels of aggression than either the extremely obnoxious one or the absence of pollution. However, contrary to a second hypothesis, pollution facilitated aggression only in the model's absence. The confirmation of the first hypothesis was interpreted as supporting and extending the affect-aggression model from heat research. It was speculated that disconfirmation of the second hypothesis resulted either from a ceiling effect or from malodor distracting attention away from the aggressive cues of the model. The methodological, theoretical, and applied implications of research on air pollution were discussed.