Fifteen men and women (six men) between the ages of 50 and 73 years were recruited to begin keep-fit classes. They were matched for sex, age, menopausal status and mass to 15 non-exercising controls. The keep-fit classes were two to three times a week and included high-impact exercise, including step and jumping exercises specifically to load the proximal femur and spine. Proximal femur, lumbar spine and total body bone mineral density (BMD) were measured at 0 and 12 months. Urinary pyridinoline (Pyr) and deoxypyridinoline (dPyr) crosslinks were measured every 6 months to assess bone resorption. Quadriceps isometric strength was measured every 6 months. BMD increased non-significantly at the femoral neck [1.57 (0.8)%] and Wards triangle [1.97 (1.4)%], and significantly at the greater trochanter 2.21 (0.9)% (P = 0.02) in the exercise group. Femoral neck BMD decreased by -1.9(0.8)% (P = 0.049) in the control group, which was significantly different from the change in the exercise group (P = 0.009). BMD did not change at the Wards triangle or trochanter in the controls. Lumbar spine BMD did not change in either group. Total body BMD did not change in the exercise group, but decreased by -0.79 (0.3)% (P = 0.02) in the controls. Following 6 months of the exercise classes. Pyr and dPyr crosslinks were significantly reduced [-19.0 (7.2)%; P = 0.0019 and -20.0 (7.7)%; P = 0.021 respectively]. There was no significant change in crosslinks after 1 year, and no change at any time in the controls. Quadriceps strength changed by 5.4 (3.7)% in the exercise group and by -6.9 (2.5)% (P = 0.01) in the control group after 12 months, being significant between groups (P = 0.008). This study suggests that high-impact, aerobic exercise in postmenopausal women and men over 50 years old is feasible and effective at maintaining muscle strength and increasing proximal femur BMD but not spine or total body BMD.