The distribution of echocardiographic left ventricular (LV) mass differs among ethnicities. Because ethnic-specific echocardiographic criteria for LV hypertrophy (LVH) are not established, we determined whether threshold values derived from overwhelmingly white populations are appropriate for blacks, a subgroup having more LVH. Between 1992 and 1994, LV mass was measured echocardiographically in the Jackson, Mississippi, black cohort of the Atherosclerosis Risk in Communities study. Participants free of prevalent cardiovascular disease (CVD) (n=1616; mean+/-SD, age 59+/-5.7; 65% women and 57% with hypertension) were included. The optimal LVH threshold value was selected from the continuum of LV mass index (LVMI=LV mass/height(2.7)) using 3 methods: (1) the best operating point from the area under the resulting receiver-operating characteristic (ROC) curve predicting incident CVD; (2) the value with the smallest probability value associated with incident CVD; and (3) visual inspection of functions of LVMI and CVD in the general additive model (GAM) plot. At a median follow-up of 6.8 years, there were 192 events (coronary heart disease=87, stroke=62, and congestive heart failure=43; incidence=17.6/1000 person-years). The best operating point from the resulting ROC analysis was 51.2 g/m(2.7) for sensitivity (53.4%) and specificity (61.5%). The Cox and GAM models adjusted for age, gender, systolic blood pressure, hypertension, diabetes, smoking, total cholesterol-to-high-density lipoprotein ratio, LVH by ECG criterion, and socioeconomic status found 50 to 51 g/m(2.7) as the optimal threshold for LVH in middle-aged blacks, corresponding to a minimum probability value and to a log-hazard ratio of zero, respectively. Because these values are close to the 51 g/m(2.7) established from predominantly white populations, this cutpoint is appropriate for both groups.