Objectives. To compare the efficacy and safety of etidronate and alendronate in patients with postmenopausal osteoporosis and to assess the efficacy of either bisphosphonate in combination with hormone replacement therapy (HRT). Patients and methods. In this pragmatic study, the main efficacy criterion was the mean annual change in bone mineral density (BMD). Patients who had a past or current history of etidronate or alendronate treatment for postmenopausal osteoporosis with at least 18 months follow-up and an evaluation in 1999 were eligible. Recruitment was in an outpatient clinic with a special focus on metabolic bone diseases. Osteoporosis was defined as at least one low-energy fracture or as a lumbar spine or femoral neck BMD decrease to at least 2.5 SD below the mean in young women. HRT was not an exclusion criterion provided treatment duration was longer than 1 year. Etidronate was given cyclically (14-day courses in a dosage of 400 mg/d separated by 76-day intervals with calcium and vitamin D supplementation) and alendronate was given daily in a dosage of 10 mg/d. Results. Of the 99 patients who met our inclusion criteria, 53 received etidronate (including 23 on HRT) and 46 alendronate (18 on HRT). Repeat BMD measurements were obtained in 88 patients, including 11 who stopped their bisphosphonate therapy within the first year of use because of adverse events. Lumbar spine BMD (mean ± SD) increased significantly both in the etidronate group (+2.1% ± 0.7%/year) and in the alendronate group (+5.3% ± 0.9%/year). The increase was significantly greater with alendronate (P < 0.01). The lumbar spine BMD increase was largest in the patients on alendronate and HRT (+6.5% ± 1.4%/year) and was smallest (and nonsignificant) in the patients on etidronate without HRT (+1.2% ± 0.8%). Femoral neck BMD showed no significant changes in any group. In the intention-to-treat analysis, fractures occurred in 12 etidronate patients (22.6%) and six (13.0%) alendronate patients (nonsignificant). Adverse events requiring bisphosphonate discontinuation before the scheduled date of the follow-up BMD measurement occurred in one patient (1.9%) in the etidronate group (generalized osteomalacia) and in ten patients (21.7%) in the alendronate group (upper or lower gastrointestinal tract symptoms in six and four patients, respectively; P < 0.01). Conclusion. Both etidronate and alendronate significantly increased lumbar BMD, but the effect was significantly more marked with alendronate. Conversely, adverse effects, most notably gastrointestinal symptoms, were more common with alendronate, so that premature treatment discontinuation because of adverse events were more common in the alendronate group. Both differences should be taken into account when selecting the best drug for a patient with postmenopausal osteoporosis.