In S.E. Kent, during 1978 and 1979, 225 patients, representing 228 occurrences of hemiplegia, were referred to a new 20-bed stroke rehabilitation ward. These patients were randomly allocated to treatment in the special ward (group A 112 strokes) or to remain in conventional treatment locations (group C 116 strokes); the two groups were closely similar, except for the initial level of consciousness. Survivors were observed at four-month intervals to one year. There are indications of benefit for patients in group A by comparison with Group C in terms of treatment received (number treated by remedial therapists, the extent of treatment, and the arrangements for aftercare), and in outcome (survival and the proportion returned to the community). The direction of the trend of advantage is consistent with another recent controlled trial, although not all modes of improvement are identical in the two trials. It is suggested that every health district should develop a comprehensive policy of stroke management.