Population-based data from the Framingham Heart Study have served as the basis for adjusting electrocardiographic (ECG) criteria for echocardiographically determined left ventricular hypertrophy (LVH) for two determinants of their sensitivity: body mass index and age. Estimated regression equations that predict echo-derived left ventricular mass from an ECG LVH voltage criterion, body mass index, and age, for 1,468 men and 1,883 women, provide a simple and effective means of adjusting that criterion for these variables. The authors evaluated five different ECG LVH criteria, comparing the performances of their original and adjusted versions within this database. All adjusted criteria significantly outperformed their unadjusted counterparts. Of these five criteria, the Cornell voltage duration product, (RAVL + SV3). QRS interval, exhibited the greatest sensitivity at all levels of specificity for both sexes (39 and 51% sensitivity at 95% specificity in men and women, respectively). Its performance was further evaluated with separate adjustment algorithms developed for lean versus obese and normotensive versus hypertensive men and women. Age and body mass index adjustment produced significant improvements for both lean and obese women and for obese men. A marginal gain in sensitivity was found in lean normotensive men. Within the relatively small subgroup of lean hypertensive men, no improvement was observed. These results suggest that among Caucasian adults, the Cornell voltage duration product adjusted for body mass index and age offers significant improvement for the detection of echocardiographically determined LVH in all but lean men; within the latter group, it loses no sensitivity in comparison with the original criterion.