OBJECTIVES: To compare health benefits and costs associated with performing bone densitometry for all men with those of risk-stratifying using the Osteoporosis Self-Assessment Tool (OST) and performing bone densitometry only for a high-risk group. DESIGN: A decision analytical model was developed using a Markov process. Three strategies were compared: no bone densitometry, selective bone densitometry using the OST, and universal bone densitometry. Data sources were U.S. epidemiological studies and healthcare cost figures. SETTING: Hypothetical cohort. PARTICIPANTS: Community-dwelling 70-year-old U.S. white men with no history of clinical osteoporotic fractures. INTERVENTION: Five years of alendronate therapy for those diagnosed with osteoporosis. MEASUREMENTS: Life years, quality-adjusted life years (QALYs), costs, and incremental cost-effectiveness ratios. RESULTS: Selective bone densitometry using the OST would cost $100,700 per additional life year gained compared to the no bone densitometry strategy. Universal bone densitometry would cost $483,500 for additional life year gained compared to selective bone densitometry. When quality of life was considered, both strategies became approximately 15% more cost-effective. Compared with the no bone densitometry strategy, selective bone densitometry would be cost saving for those aged 84 and older, with a reduction of alendronate price (< or =$110 per year), or with a higher efficacy of alendronate (a relative risk reduction of nonvertebral fracture > or =82%). CONCLUSION: Universal bone densitometry for 70-year-old men is not a good investment for society. It is reasonably cost-effective to risk-stratify with the OST, perform bone densitometry only for high-risk group, and then give men diagnosed with osteoporosis generic alendronate.