Theory. The present study examines the processes underlying the disinhibition of the eating behaviour of restrained eaters following negative emotions. Based on Herman and Polivy's (1984) Boundary Model and Wegner's Ironic Process Theory (1994), the limited capacity hypothesis is formulated, suggesting that overeating in restrained eaters results from cognitive capacity limitations. Predictions were that (1) impairment of cognitive capacity during eating will lead to overeating in restrained but not in unrestrained eaters and that (2) this difference should only emerge with food perceived to be high in calories.METHOD: The hypotheses were tested in an experiment with a 2 (restrained/unrestrained) x 2 (distraction yes/no) x 2 (perceived calories high/low) design, in which subjects consumed ice-cream in a taste test situation. Ice-cream consumption was the dependent variable.RESULTS: A second-order interaction was found: as predicted, in the high calorie condition restrained eaters ate the same amount as unrestrained eaters when not distracted, but considerably more when distracted. There was also an unexpected main effect of distraction, which indicated that restrained as well as unrestrained eaters ate more if distracted than if not distracted.DISCUSSION: The restraintxdistractionxperceived calories interaction can be explained by both the Ironic Process Theory and the Boundary Model; and the limited capacity hypothesis appears to be confirmed. The overall main effect of distraction remains puzzling. Two speculative views for the latter effect are offered.