The objectives of this study were to learn how hip fracture patients fall, and to compare the mechanics of their falls with those falls that did not result in hip fracture. In this way we sought to obtain reliable insight into the etiology and pathogenesis of hip fracture and fracture prevention. A total of 206 consecutive patients with fresh hip fracture and 100 controls were interviewed and examined between October 1994 and May 1996. The only inclusion criterion was that the fracture had occurred within 24 hours of hospital admittance. The control subjects were admitted from the same community after an accidental fall that did not result in hip fracture. The characteristics of the accident were determined by personal interview and examination of the patients within 24 hours of the event. In 98% of the hip fracture patients, the fracture was a result of a fall. The majority of the patients (76%) reported that they had fallen directly to the side. Forty-eight fracture cases had one or more eyewitnesses and their reports supported this observation. In 56% of the hip fracture patients, a fresh subcutaneous hematoma was seen on the greater trochanter of the proximal femur; such a hematoma was rare in the controls (6%) (P < 0. 001), and this gave evidence for the direct impact of the greater trochanter during the fall of the hip fracture subjects. Most of the elderly fallers who fractured a hip did not manage to break the fall, e.g., with an outstretched arm. In conclusion, our results suggest that a typical hip fracture is the result of a fall and a subsequent impact on the greater trochanter of the proximal femur. The clinical implication of this finding is that effective prevention of hip fractures could be achieved by the diminution of the number and severity of falls of the elderly. We suggest that the severity of the falls (impacts on the greater trochanter) could be decreased by an external hip protector.