Background and Purpose-Prediction of stroke in healthy individuals is challenging and there is a diurnal variation of stroke onset. We hypothesized that heart rate variability with a focus on nighttime heart rate variability will predict the risk of stroke in apparently healthy middle-age and elderly subjects. Methods-The population-based cohort of the Copenhagen Holter Study, consisting of 678 healthy subjects between age 55 and 75 years with no history of cardiovascular disease or stroke, was evaluated. All underwent 48-hour ambulatory electrocardiogram monitoring. The SD of normal-to-normal RR intervals (SDNN) was selected as the method of measuring heart rate variability. Nighttime SDNN was measured between 02:00 and 02:15 AM and could be evaluated in 653 subjects. Median follow-up was 76 months. Results-Nighttime SDNN was lower in women than in men (Pϭ0.0008), and in diabetics than nondiabetics (Pϭ0.03).However, smoking, cholesterol, systolic blood pressure, and age were not associated with nighttime SDNN.