The present study evaluates the response comparability between 361 elderly hip fracture patients admitted from the community to seven Baltimore area hospitals between 1984 and 1986 and interviewer selected proxies on items pertaining to patients' pre-fracture health and functional status. Agreement across items ranges from very poor to good and varies with respect to the health or functional area assessed. Proxies tend to overestimate patient disability relative to the patients themselves, especially with regard to capacity to perform instrumental activities of daily living. Although proxies who report the greatest contact with patients respond most comparably to the patients, when they do disagree, proxies with the greatest patient contact tend to overestimate patient disability. The authors suggest that attention to item construction and phrasing may improve response comparability.