The present study was designed to evaluate the potential role of plasma cholesteryl ester transfer protein (CETP) activity in the regulation of high-density lipoprotein (HDL) subclasses in non-insulin-dependent diabetes mellitus (NIDDM). We studied 45 men with NIDDM and angiographically defined coronary artery disease ([CAD] DM(+)CAD(+), aged 54.4 +/- 6.1 years, mean +/- SD); 47 nondiabetic men with similarly proven CAD (DM(-)CAD(+), aged 54.9 +/- 6.6 years); 43 men with NIDDM but no CAD (DM(+)CAD(-), aged 55.2 +/- 7.3 years); and 29 nondiabetic men without CAD (DM(-)CAD(-), aged 53.2 +/- 5.3 years). The groups were matched for age and body mass index (BMI). Plasma CETP activity was determined by measuring the ability of the plasma sample to transfer esterified cholesterol from exogenous C-14-cholesteryl ester-labeled low density lipoprotein (LDL) to exogenous HDL. Plasma lipoproteins were separated by ultracentrifugation. The concentration of HDL cholesterol was reduced in the DM(+)CAD(+) group as compared with the DM(-)CAD(-) group (P < .01). This change was due to a decrease of both HDL(2) cholesterol (P < .05) and HDL(3) cholesterol (P < .001). There was a clear-cut decrease in HDL(3) cholesterol in the DM(-)CAD(+) (P < .01) and DM(+)CAD(+)(P < .05) groups as compared with the DM(-)CAD(-) group. Plasma CETP activity was lower in the DM(+)CAD(-) group (1.06 +/- 0.24 arbitrary units [AU]) than in the DM(-)CAD(-) group (1.19 +/- 0.26 AU, P < .05). In the DM(+)CAD(+) group, the mean of CETP activities was 1.09 AU. Plasma CETP activity had no consistent relationship with HDL cholesterol in any of the patient groups. However, plasma CETP activity correlated with the very-low-density lipoprotein (VLDL) cholesteryl ester to triglyceride ratio in the combined group (r = +.33, n = 142, P < .001). Plasma CETP activity also correlated with total cholesterol (r = +.375, P = .011) and with LDL cholesterol (r = +.453, P = .002) in the DM(+)CAD(+) group. We suggest that plasma CETP activity is lower in patients with NIDDM as compared with nondiabetic control subjects but may not vary with respect to CAD.