BACKGROUND: the objective of this study is to investigate the detection rate of undiagnosed atrial fibrillation (AF) with short intermittent ECG recordings during four weeks among out-of-hospital patients, having at least one additional risk factor (CHADS2) for stroke. METHOD: Design: Cross-sectional study. Setting: Eight family practice centres and two hospital-based out-patient clinics in Sweden. Subjects: 989 out-of-hospital patients, without known AF, having one or more risk factors associated with stroke (CHADS2). Interventions: All individuals were asked to perform 10-second handheld ECG recordings during 28 days, twice daily and when having palpitations. Main outcome measures: Episodes of AF on handheld ECG recordings were defined as irregular supraventricular extrasystoles in series with a duration of 10 seconds. RESULTS: 928 patients completed registration. AF was found in 35 of 928 patients; 3.8% (95% confidence interval [CI] 2.7--5.2). These 35 patients had a mean age of 70.7 years (SD +/- 7.7; range 53--85) and a median CHADS2 of 2 (range 1--4). CONCLUSIONS: Intermittent handheld ECG recording over a four week period had a detection rate of 3.8% newly diagnosed AF, in a population of 928 out-of-hospital patients having at least one additional risk factor for stroke. Intermittent handheld ECG registration is a feasible method to detect AF in patients with an increased risk of stroke in whom oral anticoagulation (OAC) treatment is indicated.