This paper investigates the reliability of six measures of impairment and disability related to mobility after stroke: the Rivermead Motor Assessment (RMA, gross function subsection); gait speed (over 5 and 10 m); the motricity index (leg scores only); functional ambulation categories; sitting to standing (by observation); and mobility categories. Twenty-five patients who had suffered a stroke 2-6 years earlier leaving them with mobility disability were seen as part of a home-based physiotherapy trial. Assessments were made by three people on three occasions over 5 weeks. All six measures were reliable in statistical terms. A variation in gait speed of up to 25% and a difference of 3 points in the RMA were the actual limits of reliability.