Background. Patient preferences and expert-generated clinical practice guidelines regarding treatment decisions may not be identical. The authors compared the thresholds for antithrombotic treatment from studies that determined or modeled the treatment preferences of patients with atrial fibrillation with recommendations from clinical practice guidelines. Methods. Methods included MEDLINE identification, systematic review, and pooling with some reanalysis of primary data from relevant studies. Results. Eight pertinent studies, including 890 patients, were identified. These studies used 3 methods (decision analysis, probability tradeoff, and decision aids) to determine or model patient preferences. All methods highlighted that the threshold above which warfarin was preferred over aspirin was highly variable. In 6 of 8 studies, patient preferences indicated that fewer patients would take warfarin compared to the recommendations of the guidelines. In general, at a stroke rate of 1% with aspirin, half of the participants would prefer warfarin, and at a rate of 2% with aspirin, two thirds would prefer warfarin. In 3 studies, warfarin must provide at least a 0.9% to 3.0% per year absolute reduction in stroke risk for patients to be willing to take it, corresponding to a stroke rate of 2% to 6% on aspirin. Conclusions. For patients with atrial fibrillation, treatment recommendations from clinical practice guidelines often differ from patient preferences, with substantial heterogeneity in their individual preferences. Since patient preferences can have a substantial impact on the clinical decision-making process, acknowledgment of their importance should be incorporated into clinical practice guidelines. Practicing physicians need to balance the patient preferences with the treatment recommendations from clinical practice guidelines.