This study examines the validity of proxy assessments as substitutes for patient assessments of patient physical and psychosocial health status. Data were obtained from 275 patient-proxy pairs who were enrolled in a national study of Adult Day Health Care. Patients and proxies (informal caregivers such as spouses) were asked to complete the Sickness Impact Profile (SIP) based on the patients health status. Findings showed that patient-generated and proxy-generated physical scores were highly correlated, although proxies rated patients as slightly more impaired than the patient's rated themselves. The correlation between psychosocial scores was not high enough to consider proxy responses as valid substitutes for patient responses. We explored these differences in response by comparing regression equations predicting patient-generated and proxy-generated physical and psychosocial SIP dimension scores. Variance in the patient-generated psychosocial score was explained by physical function, psychological distress, cognitive status and patient age. Proxy-generated psychosocial scores were primarily explained by the caregiver's psychological distress and perceived burden. These findings point out the importance of considering the source of patient health status estimates when interpreting the results of research studies.