We hypothesized that the care provided to elderly patients with hip fracture in community hospitals has changed since the implementation of prospective payment systems (PPS) in 1983. We reviewed records of elderly patients admitted with hip fracture to a large community hospital from 1981 to 1986. During that period, the mean length of hospitalization decreased (from 21.9 to 12.6 days; P less than 0.0001), inpatient physical therapy decreased (from 7.6 to 6.3 sessions; P less than 0.04), and the maximal distance walked before discharge fell (from 27 to 11 m [93 to 38 ft]; P less than 0.0001). Concomitantly, the proportion of patients discharged to nursing homes rose (from 38 to 60 percent; P less than 0.0001), as did the proportion remaining in nursing homes one year after hospitalization (from 9 to 33 percent; P less than 0.0001). Neither in-hospital mortality nor one-year mortality changed significantly. As compared with beneficiaries of conventional Medicare after the implementation of PPS, HMO enrollees had shorter hospitalizations (7.3 vs. 14.0 days; P less than 0.0001), received less physical therapy (3.5 vs. 7.1 sessions; P less than 0.0001), walked shorter distances at discharge (3 vs. 13 m [11 vs. 44 ft]; P less than 0.01), and were more frequently transferred to nursing homes (83 vs. 55 percent; P less than 0.01). One year later, however, fewer HMO patients remained in nursing homes (16 vs. 35 percent; P less than 0.07). We conclude that since the implementation of PPS, hospitals have reduced the amount of care given to patients with hip fracture and have shifted much of the rehabilitation burden to nursing homes. The increase in the number of such patients remaining in nursing homes one year after the fracture suggests that the overall quality of care for these patients may have deteriorated.