Objective: To evaluate the feasibility and concurrent validity of a new, mail-administered, stroke-specific outcome measure, the Stroke Impact Scale (SIS). Design: Observational cohort study. Setting and patients: Stroke patients who had lived independently in the community prior to their stroke and who were candidates for post-stroke rehabilitation were recruited from nine, high-volume, Department of Veteran Affairs Medical Centers. Methods: Two hundred and six patients were mailed the SIS after a six-month post-stroke telephone interview. Telephone assessments included the Functional Independence Measure, the Lawton IADL and the SF-36. Results: The response rate for the mailed SIS was 63%, with 45% of the responses from proxies. The average rate of missing item level scores per patient was 1.3 (range 0-20) resulting in an average rate of 0.13 missing domain scores per patient (range 0-3). Nonresponders to the mailed SIS had more severe strokes with lower functional status at the time of the survey than responders. Proxies were more likely to complete the survey if the subjects were older, married, cognitively impaired and more functionally limited. The SIS did not exhibit a high rate of floor and ceiling effects, particularly in physical function domains, as did the FIM and the SF-36. Conclusions: The mailed SIS is a feasible means of assessing post-stroke function. Missing items and missing domain scores were extremely low, however, there is a trade-off between the low-cost mail SIS survey on the one hand and the resulting nonresponse bias on the other.