Rats exposed to a cloth impregnated with the odor of a cat made fewer contacts with the cloth and spent more time sheltering under the food and water hopper than did rats exposed to a neutral odor. In two experiments there was little evidence of between-trial habituation of these responses to cat odor. The pattern of within-session changes depended on the trial duration (15 or 60 min) and the initial level of the avoidance responses. In order to test for conditioned generalization of the avoidance responses to the test situation, rats with two previous exposures to cat odor were given a third trial in which they were exposed to a neural odor. Experiment 1 showed that those previously exposed to cat odor for two 15 min trials spent more time sheltering when exposed to the neutral odor cloth than those previously exposed to the neutral odor. Experiment 2 confirmed this effect and found that the increase in sheltering was even more marked for a group exposed to cat odor for 60 min on trial 1. The number of contacts with the neutral odor cloth on trial 3 was reduced only in the group of high avoiders (defined as making no contacts with the cat odor cloth in the first 5 min of trial 1). Thus, both the duration of exposure to cat odor and the initial response level were important in determining the conditioned generalization of the responses to a phobic stimulus.