Using Hospital Activity Analysis (HAA) data as a diagnostic index followed by record linkage procedures, a retrospective survey was carried out of patients aged 65 and over from the Newcastle area admitted to hospital with fractures of the proximal femur. Annual incidence rates for the Newcastle area of 5.6 per 1000 in females and 2.3 per 1000 in males were observed. These rates are probably 7% lower than true rates owing to patients with multiple fractures being allocated to diagnostic codes other than fractured proximal femur in HAA files. Incidence rates increased steeply with age, and rates in the sexes tended to converge at higher ages. Proportionately large numbers of patients were admitted in the winter months. The mean length of hospital stay was 75.2 days of which an average of 47.8 were spent in acute orthopaedic units: this was equivalent to the continuous occupation of 51.1% of acute orthopaedic beds in Newcastle. At one hospital, mean length of stay was 64% greater than at the other and the weekly discharge rates were suggestive of partly prescriptive discharge. Of the patients studied 35.6% died in hospital, and actuarial analysis shows that risk of death fell from initially high values to a nadir at four to six weeks and then showed a secondary rise.