Individuals tend to hold on to initial impressions even after the data upon which they have formed the impression have been discredited. To partially replicate and extend a previous study (Ross et al., 1977) on belief perserverance, in which subjects were told they were to act as clinical psychologists attempting to understand and predict patients' behavior based on background information, college students attempted to make clinical judgments about possible patient behavior based on case history material. Two control groups, one without explanation (N=20) and one with explanation (N=20) were included. Subjects in the experimental conditions provided one explanation (N=20) or two explanations (N=38) about why the patient might or might not have committed suicide. All subjects rated the likelihood of five events (committed suicide, participated in a dangerous medical experiment, became an alcoholic, contributed money to the Peace Corps, volunteered on a political campaign) subsequently occurring in the patient's life. Analysis of the results provided clear support for the hypothesis that once having explained an event, subjects' subjective likelihood predictions for that event actually occurring increase. However, no increase was obtained among subjects in the dual explanation task. It is possible that subjects recognized the ambiguities in the data when faced with a second and contradictory explanation task. (MCF)