BACKGROUND. Elderly people are prescribed more drugs than younger people. The consequences of excessive or unwise prescribing, such as drug interactions, are well known. AIM. A longitudinal study was undertaken to examine levels and patterns of prescribed drug use among a group of elderly people. METHOD. Use of prescribed drugs by a sample of elderly people in Nottingham aged 65 years and over was assessed on two occasions four years apart, in 1985 and 1989. RESULTS. Complete drug data were available for 1003 respondents in 1985 and 662 respondents in 1989 (with attrition due mainly to mortality). Drug use increased significantly with age. Women took significantly more drugs than men. Approximately half of respondents were taking at least two drugs. The overall number of drugs per person being taken in 1989 was significantly greater than in 1985. This difference remained significant when age and mortality were controlled, suggesting that changes in drug use over time within this sample may reflect genuine changes in prescribing practice (rather than simply the effects of ageing). The most commonly prescribed drug classes on each occasion were drugs for the cardiovascular system, central nervous system, musculoskeletal system, gastrointestinal tract and respiratory system. The subgroups of drugs most commonly reported at each interview were diuretics, hypnotics and anxiolytics, analgesics and non-steroidal anti-inflammatory drugs. Drugs within the category 'hypnotics and anxiolytics' showed clear and differential trends over time, with the use of anxiolytics falling, while the use of hypnotics increased. Among those respondents admitted to residential care during the course of the study higher levels of psychotropic drug use, and an increase in antipsychotic medication, were observed. CONCLUSION. It is important that the drug regimens of elderly people are frequently reviewed to ensure that only the minimum number of effective drugs, in the simplest regimen, are prescribed.