Fifty-four admissions to the psychiatric unit of a general hospital were asked to predict their length of hospitalization. The difference between their predicted and actual length of hospitalization was calculated (expectancy-reality discrepancy-ERD). Patient ERD scores were compared with self-report, ward, and therapist measures of patient improvement. Those patients who had the largest discrepancy between their expected and actual length of hospitalization (the largest ERDs) tended to improve the least while hospitalized. Improvement did not appear to be related to whether a patient was discharged sooner or remained longer than he expected. (Author)