The Barthel Index is a valid measure of disability. In this study we investigated the reliability of four different methods of obtaining the score in 25 patients: self-report, asking a trained nurse who had worked with the patient for at least one shift, and separate testing by two skilled observers within 72 hours of admission. Analysis of total (summed) scores revealed a close correlation between all four methods: a difference of 4/20 points was likely to reflect a genuine difference. In individual items, most disagreement was minor and involved the definition of middle grades. Asking an informed nurse or relative was as reliable as testing, and is quicker.