Health screening for old people who live at home has been the subject of debate for 30 years or so. It has come to the fore again in the UK with the new emphasis on annual assessments by general practitioners (GPs) of those aged 75 or more. Screening in the elderly has implications for manpower. How can it best be done? We describe here a randomised, controlled study of case finding and surveillance in patients aged 65 and over in a general practice in South Wales. Problem identification was by a postal questionnaire, focusing on function, that was sent at random to 369 eligible patients with subsequent verification and intervention by a specially appointed nurse. The 356 controls had no questionnaires and no contact with that nurse. The study lasted 3 years, and end-points included mortality, self-ratings of quality of life, and health status, and use of all services (GP contacts, hospital admission, home help, and so on). Mortality was significantly lower in the intervention group (18%) than in the controls (24%) (difference 6.0% [95% CI 0.1-11.9%], p less than 0.05). Total number of hospital admissions did not differ between intervention and control groups, but duration of hospital stay of patients aged 65 to 74 years was significantly shorter in the intervention group (difference 4.6 days [95% CI 1.6-7.6], p less than 0.01). An increase in visits to a GP was largely offset by a lower number of home visits by a GP. Quality-of-life measures revealed no between-group differences, but self-rated health status was superior in the intervention group. We conclude that the use of a postal screening questionnaire with selective follow-up and intervention can favourably influence outcome and use of health care resources by elderly people living at home.