The effect of hypnotic use on self-rated quality of sleep and therapist-rated level of alertness was examined in an inpatient rehabilitation setting. We examined what other factors were predictive of a restful sleep in this population. Seventy-five inpatients at the Jewish Rehabilitation Hospital in Montreal were included. Patients were asked to rate the quality of their own sleep on a given night. Night nurses recorded whether sleeping pills had been used and rated patients' sleep and number of awakenings during the same night. Patients were evaluated by their physiotherapists and occupational therapists the next day regarding how well rested they seemed according to three parameters: alertness, fatigue, and level of participation in therapy. Thirty-three percent of the patients received sleeping pills on the study night. Sleeping pill use did not predict patient perception of getting a good night of sleep or the somewhat more objective sleep rating by the night nurse. Whether a sleeping pill was taken was also found not to be predictive of restful sleep as estimated by the physical and occupational therapists. Variables significantly associated with therapists' ratings of apparently restful sleep included number of comorbidities, the nurses' rating of how well the patient had slept, the patients' self-assessment of sleep, and whether the patient felt well rested the morning after sleep. However, the patients' own assessment of sleep quality was negatively related to their performance in rehabilitation therapy. This suggests that patient self-report of sleeping difficulty may not be the best or only guideline to follow when considering intervention such as prescribing sleeping pills, particularly because sleeping pill use seems not to influence either patient perception of sleep or how well rested they seem in therapy.