Three studies investigated the effects of extreme context stimuli and categorization on hedonic contrast by having subjects judge the attractiveness of faces. Experiment 1 demonstrated hedonic contrast in both directions by using 2 sets of stimuli presented in different orders. Preceding moderately unattractive faces with moderately attractive faces made the unattractive faces more unattractive. When the order of presentation was reversed, the moderately attractive faces became more attractive. Experiment 2 found that this hedonic contrast was eliminated when the moderately attractive faces were replaced with extremely attractive faces. Experiment 3 showed that even with those 2 sets of extremely different stimuli, hedonic contrast occurred if subjects were instructed to think of both sets of stimuli as belonging to the same category. These findings, using hedonic judgments, parallel Sarris's (1967, 1968) finding with weights that when 2 sets of stimuli are too different in the dimension being judged, no contrast occurs. They also lend support to his explanation for this result. When the 2 sets of stimuli are too different they are not seen as belonging to the same category. They are therefore not compared, and contrast does not occur. The authors propose that these principles might apply to contrast in all settings. (Contains 2 footnotes.)