Osteoporosis is a large and growing disease category with significant health consequences. Information on the cost-effectiveness of preventing and treating this disease may help healthcare payers allocate resources more efficiently. OBJECTIVE: To assess cost-effectiveness of bisphosphonate therapies in multiple populations, subject to a value threshold. METHODS: A fracture incidence-based Markov model of osteoporosis, where patients transition across outcome states over time (e.g., fracture, healthy, dead), was used to estimate incremental cost per QALY gained ratios. The base case analysis was conducted on a cohort of women aged 65 years with low bone mineral density (BMD) and prevalent vertebral fracture, with 3 years of treatment with Actonel or Fosamax, using a 3-year time horizon. Model inputs included fracture incidence rates, relative risk (RR) reduction of fracture due to risk factors, fracture costs, prices/day (Actonel $1.95; Fosamax $2.21), health utilities, and efficacy in terms of relative risk (RR) of fracture reductions for hip (60% Actonel; 51% Fosamax) and vertebral (49% Actonel; 47% Fosamax). A 3% discount rate was applied to costs and outcomes. Multiple populations were evaluated by varying efficacy rates (upper/lower 95% confidence intervals), fracture costs (+/-25%), utility values (+/-50%), fracture RR (2.0 - 7.0), age (55-75), therapy discontinuation (0% - 76%), and time horizon (to lifetime). RESULTS: Using a $30,000 per QALY gained cost-effectiveness threshold, Actonel compared to no treatment was cost-effective in the majority of populations. Under base case assumptions Actonel's cost/QALY gained was $16,158 and dominated Fosamax (i.e., less costly and more effective). Actonel's cost/QALY ratio crossed the threshold as RR of fracture declined (<5), starting cohort age fell (<62), discontinuation rose, fracture costs decreased (>25%), or fracture-related utility decrements decreased (>50%). CONCLUSIONS: Many high-risk patient subpopulations can be treated economically with bisphophonates. The most significant drivers are RR of fracture, starting cohort age and time horizon.