The purpose of this study was to compare outcomes in stroke patients with two different methods of admission to an inpatient unit. Patients' outcomes were studied for 3 years. Admission screening was performed either by a physiatrist (group I, n = 93) or by a nurse practitioner (group II, n = 100). All admission decisions were made solely by the physiatrist. The outcome measures used were appropriateness of admission, discharge to home versus nursing home and functional gains measured on the Functional Independence Measure. The average length of stay was 28 days for group 1 patients and 29 days for group II. Ninety-seven per cent of group I and 94% of group II patients were considered appropriate for admission. The difference is not statistically significant (Z = 1.02). Ninety-six per cent of patients in each group were discharged home. Ninety-six per cent of group I and 89% of group II patients met the functional goals. Group I fared better than group II, and the difference between the groups (Z = 1.85) is statistically significant at p < 0.04.