Objective: Poor adherence to therapy, perhaps related to unaddressed patient preferences, limits the effectiveness of osteoporosis treatment in at-risk women. A parallel patient-level randomized trial in primary care practices was performed. Methods: Eligible postmenopausal women with bone mineral density T-scores less than -1.0 and not receiving bisphosphonate therapy were included. In addition to usual primary care, intervention patients received a decision aid (a tailored pictographic 10-year fracture risk estimate, absolute risk reduction with bisphosphonates, side effects, and out-of-pocket cost), and control patients received a standard brochure. Knowledge transfer, patient involvement in decision-making, and rates of bisphosphonate start and adherence were studied. Data came from medical records, post-visit written and 6-month phone surveys, video recordings of clinical encounters, and pharmacy prescription profiles. Results: A total of 100 patients (range of 10-year fracture risk, 6%-60%) were allocated randomly to receive the decision aid (n = 52) or usual care (n = 48). Patients receiving the decision aid were 1.8 times more likely to correctly identify their 10-year fracture risk (49% vs 28%; 95% confidence interval [CI], 1.03-3.2) and 2.7 times more likely to identify their estimated risk reduction with bisphosphonates (43% vs 16%; 95% CI, 1.3-5.7). Patient involvement improved with the decision aid by 23% (95% CI, 13.6-31.4). Bisphosphonates were started by 44% of patients receiving the decision aid and 40% of patients receiving usual care. Adherence at 6 months was similarly high across both groups, but the proportion with more than 80% adherence was higher with the decision aid (n = 23 [100%] vs n = 14 [74%]; P = .009). Conclusion: A decision aid improved the quality of clinical decisions about bisphosphonate therapy in at-risk postmenopausal women, did not affect start rates, and may have improved adherence. © 2011 Elsevier Inc. All rights reserved.