This study tested the hypothesis that cognitive disturbance has a causal role in the maintenance of disturbed eating in bulimia nervosa. Thoughts about eating, weight and shape were activated in one group of patients with bulimia nervosa (the experimental group) but not in another (the control group). There was an increase in negative self-statements in the experimental group following the experimental manipulation. Food consumption was then measured in a taste test. As predicted, the experimental group ate less in the short term than the control group. Contrary to expectations, they did not subsequently report more objective bulimic episodes as a consequence of this decreased food intake. They did, however, report significantly fewer subjective bulimic episodes.