Low density lipoprotein (LDL) particles are heterogeneous in size, density, and chemical composition; small, dense LDL may be more atherogenic than large, buoyant LDL. We have developed a rapid microscale method called LDL VAP-II (Vertical Auto Profile-II) for quantification of cholesterol in LDL subclasses. The method is based upon a short (1 h) single vertical spin density-gradient ultracentrifugation and on-line VAP-II analyzer. LDL VAP-II is rapid and reproducible. Using this method five LDL subclasses, designated as LDL-1 (most buoyant) through LDL-5 (most dense), have been identified in a population consisting of 195 medical students (ages, 22-29 years). The R(f) (relative position of the major LDL peak in the density gradient; the higher the R(f) value, the lower the peak density) was significantly positively correlated with cholesterol levels of high density lipoprotein (HDL) (r = 0.594), HDL3 (0.350) and HDL2 (0.625), and significantly negatively correlated with triglycerides (TG) (-0.355) and cholesterol levels of very low density lipoprotein (VLDL) (-0.386) and intermediate density lipoprotein (IDL) (-0.432). These results are consistent with those obtained by other investigators. The R(f) value was significantly correlated with peak particle diameter as determined by non-denaturing gradient gel electrophoresis (r = 0.859). In a forward stepwise multivariate analysis comparing R(f) with sex, VLDL, LDL, Lp[a], IDL, HDL3, HDL2, and triglyceride, only HDL2 remained in the model.