Late potentials (LPS) have been shown to be predictive of ventricular tachycardia in coronary artery disease subjects, but the prevalence of LP in the general population is not as well-known. The study included 487 men without a history of cardiovascular disease (aged 50-59) living in Southwestern France. Standard-averaged high resolution electrocardiogram was performed using Butterworth filtering at 40-250 Hz. LPS were defined as two or more of the following criteria: QRS > 114 ms, duration of the low amplitude signals in the terminal portion of QRS > 38 ms, root mean square (RMS) voltage of the last 40 ms < 20 microV. The mean QRS duration was 97 +/- 12 ms (mean +/- SD), duration of the low amplitude signals in the terminal portion of QRS was 32 +/- 10 ms, and RMS voltage in the last 40 ms was 39 +/- 27 microV. Eight percent of subjects (95% confidence interval [CI]: 6%-11%) had a QRS duration > 114 ms; 22% (95% CI: 18%-26%) had a duration of low amplitude signals > 38 ms, and 25% (95% CI: 22%-29%) had RMS voltage in the last 40 ms < 20 microV. Finally, the prevalence of LP was 21% (95% CI: 18%-25%). In conclusion, according to commonly used criteria, the prevalence of LP in this healthy population of middle-aged men is 21%, close to the values found in the literature for myocardial infarction patients. Those findings indicate the need for reconsidering the definition of LPS.