The objective of this study was to establish the reliability and sensitivity of both postal and interviewer-administrated versions of the Rivermead Extended Activities of Daily Living (READL) index, which assesses six domestic activities and six community activities. Sixty patients with stable neurological impairment were recruited. In one group (n = 40), every patient was assessed face-to-face using the READL, the Barthel index (BI) and the short orientation memory and concentration test (SOMC). One week later, the READL was repeated by the same person, in the same place. In the second group (n = 20), all the patients were first sent a postal form of the READL and were then seen face-to-face for assessment as in group 1. To be included patients had to score at least 18/28 points on the SOMC. Scores were compared using scatterplots, Bland and Altman plots and correlation coefficients, and difference scores were calculated. Sensitivity was established comparing groups of patients expected to differ in their activities. Repeated assessment score, both face-to-face and by post, showed significant correlation (Pearson coefficient = 0.97 and 0.88, respectively). Most scores were within four points of each other, with no systematic bias, although patients tended to rate themselves more independent. Both methods were able to detect differences in the level of activities as predicted between more and less dependent groups (t-test: p < 0.00001 and p = 0.00087). The READL index appears to be a reliable and sensitive measure, with some evidence for validity, but further research is needed.