BACKGROUND: The integration of bone density testing into well-designed fracture prevention programs that can be applied in populations has not been studied. OBJECTIVES: We sought to compare the outcomes of 3 strategies for allocating bone density testing within an HMO-based fracture prevention program. RESEARCH DESIGN: Women were randomly sampled and allocated to one of 3 groups: (1) a universal group, in which all were offered bone mineral density (BMD) testing (1986 contacted; 415 participated); 2) the SCORE group, in which women scoring > or = 7 on the SCORE questionnaire were invited for BMD testing (1940 contacted; 576 participated); and (3) the Study of Osteoporotic Fracture (SOF)-based group, in which women with > or = 5 hip fracture risk factors were invited for BMD testing (5342 contacted; 2176 participated). SUBJECTS: Women aged 60-80 not taking hormone therapy or osteoporosis medication were included. MEASURES: Outcomes ascertained during 33 months of follow-up in all women contacted included initiation of osteoporosis treatment and hip and total fracture rates. Outcomes evaluated among all participants included changes in fracture risk factors, osteoporosis knowledge, and satisfaction with the program. RESULTS: Osteoporosis treatment rates did not differ among all women contacted but were slightly higher among trial participants in the universal and SCORE groups (21.1% and 20.2%, respectively; versus 16.7% in the SOF-based group (P value versus universal = 0.04). Among all women contacted, fracture rates were lowest in the universal group (74.11/1000) and differed significantly compared with the SCORE (99.44/1000; P = 0.009) and SOF-based groups (91.77/1000;P = 0.02). Knowledge about osteoporosis risk factors was highest in the universal group and lowest in the SOF-based group (P < 0.01). CONCLUSIONS: The degree to which BMD testing was offered to women in a fracture prevention program significantly affected total fracture rates, change in some fracture risk factors, and knowledge about risk factors.