Early detection of atrial fibrillation (AF) is desirable for preventing strokes. Not only does AF often go undetected in patients being followed up for various disease conditions, but the optimal detection method also remains to be elucidated. In a prospective observational study of 24-h Holter monitoring versus 14-day external loop recording performed for detection of previously undiagnosed AF in 868 Japanese outpatients (aged 75 ± 6 years), with a CHA2DS2-vasc score ≥ 1, but no prior AF episodes, AF was detected during the initial monitoring period in 16 (1.8%) patients, in 7 (1.1% [7/645]) by 24-h monitoring and 9 (4.0% [9/223]) by 14-day monitoring (P = 0.005), and overall in 32 (3.7%) during the 1-year study period. Absence of a beta-blocker therapy and the serum N-terminal pro-brain natriuretic peptide level were independent predictors of a new detection of AF. Oral anticoagulation (OAC) therapy was given to 22 (69%) of the 32 patients in whom AF was detected, and no difference in the incidence of subsequent major adverse events was found between the patients managed with and without oral OAC therapy. Previously unknown AF was documented at a prevalence of 3.7% per year among Japanese with a notable CHA₂DS₂-VASc score, and 14-day external loop monitoring was significantly more effective for detection of the disorder. A large-scale prospective AF screening study conducted to clarify the type or types of patients who would benefit from “early” OAC therapy for primary stroke prevention is warranted.