We studied the effect on bone mass of alendronate treatment for 5 yr and its withdrawal. Four hundred and forty-seven postmenopausal women with normal bone mass entered a 3-yr randomized trial followed by a 2-yr open label extension. Three hundred and eleven women completed the first 3 yr, and 263 consented to continue and completed the extension. We are reporting data from groups using the dose of alendronate currently approved for osteoporosis prevention (5 mg) or from the group in which alendronate treatment was withdrawn: 52 women received alendronate (5 mg) for 5 yr (group I), 56 received 3 yr of placebo followed by alendronate (5 mg) for 2 yr (group II), and 52 received alendronate (20 mg) for 2 yr followed by 3 yr off therapy (group III). In group I, alendronate (5 mg) increased bone mineral density (BMD) at the spine and trochanter by 2.5-3.2% (P < 0.001 vs. baseline) and stabilized total body and femoral neck BMD (change vs. baseline, P = NS) over 5 yr. By the end of 5 yr, BMD was comparable at the spine, hip, and total body in groups I and III. The 3-yr decrease in BMD after withdrawal of alendronate (20 mg) in group III was 1.8-5.7% (P < 0.01 vs. baseline) and similar to the 3-yr decrease in BMD in group II during the initial 3 yr. In conclusion, alendronate (5 mg) for 5 yr or alendronate (20 mg) for 2 yr followed by 3 yr off therapy prevented postmenopausal bone loss. After withdrawal of alendronate (20 mg), bone loss resumed at the normal early postmenopausal rate.