We previously found that dietary soy protein produces higher lipoprotein(a) [Lp(a)] plasma concentrations than does casein. This study tested the hypothesis that soy protein contains Lp(a)-raising alcohol-removable components. Twelve normolipidemic women and men consumed, in a crossover design, liquid-formula diets containing casein, soy protein, or alcohol-extracted soy protein. Dietary periods of 32 days were separated by washout periods on self-selected diets. Fasting lipid and Lp(a) levels were measured throughout. Median Lp(a) concentration was >2-fold greater after 28 to 32 days on a soy protein diet than after an extracted soy protein diet (P<0.001). Lp(a) concentrations after casein and extracted soy protein diets were virtually identical. Women and men responded similarly. When the switch was made from a self-selected to a soy protein diet, median Lp(a) concentration increased 16% after 1 week (P<0.01) and subsequently decreased toward baseline; extracted soy protein and casein diets never exhibited increased median Lp(a) levels, and after 28 to 32 days, these levels were decreased >60% below baseline (P<0.001 and P<0.01, respectively). Low density lipoprotein cholesterol concentrations were not different after the 3 experimental diets. The data indicate that (1) dietary soy protein can increase Lp(a) concentrations, (2) this effect is eliminated after alcohol extraction, and (3) high Lp(a) concentrations may be markedly reduced by diet.