AIM: Our aim was to investigate the prevalence and the prognostic significance for fatal and nonfatal cerebrovascular and cardiovascular events of different ECG criteria for left ventricular hypertrophy (LVH) in normal weight, overweight and obese patients in an adult Italian population. METHODS: A total of 18 330 adults (mean age 54 ± 11 years, 55% women, 53% hypertensive patients) were analyzed from the Moli-sani cohort. Obesity was defined using the ATPIII criteria. ECG-LVH was defined according to 2013 ESC-ESH guidelines. RESULTS: The age and sex adjusted prevalence of ECG-LVH did not differ from normal weight patients to class 1-3 obesity patients, when Cornell-voltage criterion was used. In overweight and obese patients, as compared with normal weight patients, a progressively lower prevalence of ECG-LVH was observed when the Sokolow-Lyon index was used, whereas a higher prevalence was shown by using the aVL R-wave voltage (>11 and >5.7 mm) and the Cornell-voltage-QRS duration product. The incidence of cardiovascular events was significantly greater in patients with ECG LVH diagnosis by the Cornell voltage [hazard ratio 1.89, 95% confidence interval (CI) 1.05-3.39] and the Cornell product (hazard ratio 1.87, 95% CI 1.31-2.67). After adjusting for different confounders (age, sex, cigarette, hypertension, hypercholesterolemia, diabetes, income, education, occupational class and physical activity) and for BMI categories, only the Cornell product remained significantly associated with a higher incidence of cardiovascular events (hazard ratio 1.66; 95% CI 1.16-2.38). The predictive significance of different LVH criteria was assessed across BMI categories; after adjusting for confounders, no LVH criteria were significantly associated with an increased risk of cardiovascular events in obese patients; Cornell-product LVH remained an independent predictor of events in normal weight and overweight individuals (hazard ratio 2.63; 95% CI 1.10-6.28 and hazard ratio 2.72; 95% CI 1.52-4.25, respectively). CONCLUSION: Our results confirm that ECG LVH prevalence may differ according to the criteria used across BMI categories in a low cardiovascular risk cohort. The use of different LVH criteria according to BMI categories may improve cardiovascular risk stratification in a general population independently of several confounding factors.