We measured the capacity of human plasma to induce cholesterol efflux from Fu5AH rat hepatoma cells in four groups of men with or without non-insulin-dependent diabetes mellitus (NIDDM) and coronary artery disease (CAD). Plasma from men with both NIDDM and CAD (n = 47) had the lowest efflux capacity (17.3 +/- 3.6%) whereas healthy control subjects with neither diabetes nor CAD (n = 25) had the highest capacity (19.8 +/- 3.4%). The groups with CAD but no diabetes (n = 44) and with NIDDM but no CAD (n = 35) had intermediate efflux values (18.5 +/- 3.8 and 18.5 +/- 3.9%, respectively). In a 2 x 2 factorial ANOVA, the differences were significant with respect to the presence of CAD (P = 0.038) and NIDDM (P = 0.041), with no interaction between the factors. The concentration of HDL particles containing apolipoprotein (apo) A-I but no apo A-II (LpA-I) was not related to efflux capacity in univariate or multivariate analyses. A multivariate regression analysis showed that when controlled for the presence of NIDDM and CAD, the concentration of particles containing both apo A-I and apo A-II (LpA-I:A-II) and plasma phospholipid transfer protein activity were both positively, independently, and significantly (P < 0.001) related to cholesterol efflux capacity.