Reality monitoring performance was assessed in normal, manic, and schizophrenic subjects, with self-generated text as the stimuli. The subjects were asked to mentally plan or orally generate stories and then to identify the origin ofthe information that they had planned or generated. The subjects were also exposed to irrelevant, distracting story material while performing the task. The performance of all subjects was appreciably better than performance in other studies in which word-list reality monitoring was used, and distraction had no deleterious effect on performance. These results are discussed in terms of their implications for intentional versus unintentional encoding processes in schizophrenics.Reality monitoring (Johnson & Raye, 1981) is the process of attribution of the origin of information in memory to an internal or external source. In several studies of reality monitoring, Johnson and her colleagues have identified a number of characteristics of reality monitoring in normal individuals. For example , normal individuals manifest certain typical decision biases (Raye & Johnson, 1980), in that, when they make attribution errors, they tend to report that they have only thought information that they have actually said. Furthermore, reality-monitoring discrimination ability appears to develop over time (Foley, Johnson, & Raye, 1983), although decision error biases appear to be relatively stable across the developmental span.Several theorists (e.g., Maher, 1983;Rochester, 1979) have asserted that communication disturbances in schizophrenia may be caused by failures of aspects of information-processing competence similar to those subsumed by the process of reality monitoring. In earlier studies of reality-monitoring ability in psychotic patients, Harvey (1985;Harvey, Earle-Boyer, & Levinson, 1988) found that schizophrenic patients had characteristic problems in reality monitoring. Schizophrenicpatients who manifested verbal communication disturbances (i.e., " formal thought disorder") were less competent than normals at discriminating the actual origin of information that they had said, as opposed to thought, and tended to make recognition errors that were the polar opposite of normal biases. Harvey et al. (1988) found that the concurrent presence of reality-monitoring discrimination problems and a "think-report-say" recognition error bias predicted the In this report we present data from a modification of the reality-monitoring procedure as applied to schizophrenic and normal subjects and a psychiatric control sample . We changed the typical reality-monitoring task, in which the subject is asked to discriminate the origin of word stimuli from a list, by having the subjects discriminate the origin of word stimuli that had been contained in a story that they had either planned or planned and generated. We also had the subjects perform realitymonitoring operations while being exposed to irrelevant information , in order to manipulate overall demand on information-processing capacity. Schizophrenics typicall...