We examined associations between metabolic variables and changes in coronary artery disease (CAD) in the St. Thomas' Atherosclerosis Regression Study (STARS). The course of CAD over 3 years was measured continually by quantitative coronary angiography, i.e., as the per patient change in the mean absolute width of coronary segments (delta MAWS). The decrease in MAWS (progression of CAD) was significantly correlated with in-trial plasma concentrations of cholesterol (P = 0.002), low-density lipoprotein (LDL) cholesterol (P = 0.001), apolipoprotein B (apoB) (P = 0.008), and lipoprotein(a) [Lp(a)] (P = 0.004); no significant associations were found with high-density lipoprotein (HDL) cholesterol, apoA-I, vitamin E, thyroid hormones, fibrinogen, von Willebrand factor, or post-load plasma glucose and insulin concentrations. By multiple regression analysis, LDL cholesterol was the best predictor of delta MAWS, the adjusted model explaining 22% of the variance (P = 0.04). Thus, in men with symptomatic CAD the most important metabolic predictor of change in CAD is plasma LDL cholesterol, there being no advantage in measuring other variables, in particular, apoB or Lp(a).