We studied the echocardiographic measurements and the digital electrocardiograms (ECGs) of 3378 participants in the Cardiovascular Health Study and divided the subjects into a Learning Set of 1678 subjects and a demographically matched Test Set of 1700 subjects. We defined left ventricular hypertrophy (LVH) as values of the echocardiographic left ventricular mass index (LVMI) >116 g/m(2) in men or >104 g/m(2) in women. We performed univariate linear regression separately on the male and female subjects in the Learning Set to find all the ECG parameters that correlated significantly with LVMI. We then applied multivariate linear regression (MLR) to these parameters to identify the 4 variables for each sex that discriminated best between the subjects with and without LVH. We then applied the resultant sex-specific linear regressions to the subjects in the Test Set. We found that the diagnostic performances for LVH of the new MLR equations were superior to those of three commercial computerized ECG diagnostic algorithms in both men and women. We also found that for the new MLR and for all three commercial algorithms, the diagnostic sensitivities at matched specificities were superior in the men compared to those in the women. We conclude that the new MLR equation identifies patients with echocardiographically determined LVH more reliably than do 3 commercial computerized diagnostic algorithms and that the sensitivities of each of these modalities is greater in men than it is in women.