Three experiments investigating the role of boredom as a limiting condition on Zajonc's (1968) mere exposure effect are described. In Experiment 1, non-boredom-prone Ss showed significant exposure effects for Welsh figure stimuli, whereas boredom-prone Ss showed no exposure effects at all for these stimuli. In Experiments 2 and 3, complex stimuli (line-drawn optical illusions) produced significantly stronger exposure effects than relatively simple stimuli (Welsh figures). The difference in affect ratings for optical illusion vs. Welsh figure stimuli was greater when Ss rated both types of stimuli (Experiment 2) than when Ss rated only 1 type of stimulus (Experiment 3). Furthermore, Welsh figures showed a decline in affect ratings with increasing exposure frequency in Experiment 2 and an increase in affect ratings with increasing exposure frequency in Experiment 3, suggesting that stimulus "contrast" effects are important in determining affect judgments in mere exposure experiments. Results support the role of boredom as a limiting condition on the mere exposure effect and are consistent with a 2-factor learning-satiation model of the exposure effect.