The prevalence and incidence of CHD, defined by ECG abnormalities according to the Tecumseh criteria for Minnesota Codes, were determined in Pima Indians greater than or equal to 25 yr of age. In a cross-sectional analysis, the age-sex-adjusted prevalence (+/- SE) of ECG abnormalities was higher in 1454 NIDDM patients (6.86 +/- 0.65%) than in 1696 nondiabetic subjects (3.23 +/- 0.63%; prevalence rate ratio = 2.12; 95% CI 1.39-3.25). In a prospective analysis, the age-sex-adjusted incidence (+/- SE) of ECG abnormalities was higher in 824 NIDDM patients (12.77 +/- 1.67) than in 935 nondiabetic subjects (5.93 +/- 1.43 cases/1000 person-yr; incidence rate ratio = 2.15; 95% CI 1.26-3.69). The prevalence of ECG abnormalities in insulin-treated NIDDM patients was significantly higher than in NIDDM patients not treated with insulin (age-sex-adjusted OR = 2.83; 95% CI 1.84-4.33); and this association persisted when adjusted for other factors such as sBP, BMI, duration of diabetes, serum cholesterol concentration, and oral hypoglycemic agents (OR = 2.12; 95% CI 1.34-3.37). In the prospective analysis, the incidence of ECG abnormalities in NIDDM patients treated with insulin was higher than in those NIDDM patients not treated with insulin, but, when controlled for age, sex, duration of diabetes, and oral hypoglycemic agents in a proportional-hazards model, the relationship with insulin treatment was not statistically significant (incidence rate ratio = 1.36; 95% CI 0.80-2.31). This suggests that insulin treatment may be a marker of more severe diabetes, and that factors associated with clinical indications for insulin treatment, rather than insulin treatment per se, are related causally to CHD. On the other hand, endogenous fasting and 2-h postload serum insulin concentrations were not associated with ECG abnormalities among 761 NIDDM patients not treated with insulin nor among 1226 nondiabetic subjects. Furthermore, in the prospective study, neither endogenous fasting nor 2-h postload serum insulin was associated with the subsequent development of ECG abnormalities in NIDDM patients or nondiabetic subjects.