This study aimed to establish the inter-rater reliability of a novel measure of stroke outcome which has been developed for use in large randomised trials and which could be used in any large follow-up study. One hundred and thirty-two stroke patients were asked the two ''simple questions'' during a face-to-face interview on two occasions approximately 6 months after their stroke. The inter-rater reliability of the ''dependency'' question (i.e. Do you require help from another person with everyday activities?) was moderate ((= 0.51) whilst that of the ''recovery'' question (i.e. Do you feel that you have made a complete recovery from your stroke?) was good ĸ= 0.61). Although this degree of reliability may be acceptable in very large studies, it will make it more difficult to demonstrate real differences between groups. We discuss the possible reasons for the moderate reliability and suggest modifications to the questions which we hope further testing will show improves their performance