Antibiotics are being proposed for the treatment of cardiovascular disease. In the past, antibiotics were advocated for the control of hypercholesterolemia. We have therefore investigated the relation between colonic bacterial activity and serum lipids. In a four-phase randomized crossover study, we fed a different starch supplement during each 2-week phase to 24 healthy subjects. In two phases, supplements containing resistant starches were fed that reach the colon and are largely fermented by colonic bacteria. Fecal starch recovery therefore reflects the metabolic activity of colonic microflora. The control treatments were conventional starches. Blood lipid levels were obtained at the start and 4-day fecal collections at the end of each phase. Resistant starch supplements increased fecal starch excretion by 3.8 +/- 1.2 g/d more than conventional starches (P = .006). Mean starch excretion was related positively to pretreatment serum high-density lipoprotein (HDL) cholesterol (r = -.57, P = .003) and negatively to low-density lipoprotein (LDL) cholesterol (r = -.57, P = .004), apolipoprotein B:AI (r = -.56, P = .005), and fecal output of fusobacteria (r = -.73, P = .003) and bacteroides (r = -.72, P = .003). The ratio of fusobacteria to total anaerobes was also related to pretreatment LDL cholesterol (r = .56, P = .037). Differences in starch excretion between healthy subjects, as a measure of bacterial activity, accounted for 32% of the variation in pretreatment LDL cholesterol. The activity of colonic microflora therefore appears to influence serum lipid levels. Alterations of bacterial number and activity may provide an additional strategy to control serum lipid risk factors for cardiovascular disease.