BACKGROUND: Heart rate variability (HRV) is usually reduced in patients with CAD. We therefore investigated whether reduced HRV is predictive of angiographic CAD beyond Framingham risk in patients with stable angina. METHODS AND RESULTS: A total of 514 patients (age, 66.1 +/- 14.3 years, 358 men) were enrolled. Holter ECG was performed before catheterization, and 24-h HRV was analyzed in both the frequency domain (VLF, LF, HF and total power) and the time domain (SDNN, SDANN, RMSSD and pNN20). Angiographic CAD was defined as >/= 50% diameter reduction of 1 or more coronary arteries. On coronary angiography 203 patients (39.6%) had angiographic CAD. Patients with CAD had significantly higher Framingham risk and lower HRV according to both frequency and time domain parameters. After controlling for age, gender, heart rate, SBP, renal function, lipids and Framingham risk, reduced HRV indices remained predictors of CAD (OR, 95% CI for LF, HF, SDNN, RMSSD and pNN20: 0.81, 0.66-0.99; 0.77, 0.63-0.94; 0.75, 0.59-0.96; 0.72, 0.58-0.88; and 0.76, 0.62-0.94, respectively). On subgroup analysis, HRV parameters appeared to be predictive of CAD only in subjects with high Framingham risk or diabetes. CONCLUSIONS: Reduced HRV is predictive of CAD in patients with stable angina, independent of traditional risk factors and Framingham risk. The predictive value of HRV may be relevant only in subjects with high Framingham risk or diabetes.