The aim of the study was to examine the role of peak bone mass and rate of postmenopausal bone loss for the subsequent risk of osteoporotic fracture. 182 women within 3 years of menopause were followed longitudinally for 15 years. Over the first 2 years, forearm bone mass (single photon absorptiometry) was measured nine times, the rate of bone loss was calculated, and the women were stratified into a group of "fast bone losers" (n = 49) and a group of "normal bone losers" (n = 133). Later, bone mass was also measured in the lumbar spine and hip with dual energy X-ray absorptiometry. At 15 years, the fast losers had significantly lower body weight (4.4 kG; p < 0.05) than the normal losers. Furthermore, the fast losers had significantly increased values of bone turnover (osteocalcin and C-terminal type I collagen breakdown products). In both the forearm, spine, and hip, the fast bone losers had at all sites significantly less bone mass than the normal bone losers (p < 0.001). 23 women had experienced a peripheral (Colles') fracture and 25 a spinal fracture. The fracture groups had generally significantly (p < 0.05) less bone mass than the group without fracture, both in the forearm, spine, and hip and they also had the highest rate of bone loss after menopause (p < 0.05). Baseline bone mass and rate of loss predisposed to the same extent to fractures with ODD's ratios of about 2. If both low bone mass and rate of loss were present, the ODD's ratio increased to about 3. We conclude that fast rate of bone loss and low bone mass are equally important for the risk of fracture. The identification of women at risk of osteoporosis should therefore consider both a measurement of bone mass status, and a determination of the postmenopausal rate of loss.