A 4% random sample of elderly (i.e. 65 years and over) was followed up for 3 months after their discharge from N.H.S. non-psychiatric hospitals throughout Wales. Seventeen per cent were readmitted within 3 months of their original discharge. The proportion of readmissions did not demonstrate any significant statistical association with social or demographic characteristics of patients. Rather the study suggested that readmissions were due to a relapse or breakdown of the original medical condition.