Beta-casein is a cow's milk protein that occurs predominantly in two forms, A1 and A2. Epidemiological evidence suggests that per capita consumption of beta-casein A1 is associated with national mortality rates from ischaemic heart disease. A biological mechanism was proposed after rabbits fed diets containing beta-casein A2 had lower serum cholesterol concentrations and less aortic intimal thickening than rabbits fed beta-casein A1. We tested whether beta-casein A1 and A2 variants differentially affect plasma cholesterol concentrations in humans. In a randomised crossover trial of two four-and-a-half week periods without washout, 62 participants replaced all dairy products in their diet with 500 mL of low-fat milk and 28 g of full-fat cheese that differed in the proportion of beta-casein A1 and A2 variants. Duplicate blood samples were taken on non-consecutive days at the end of each treatment period from 55 people who completed the study. Mean (S.D.) plasma total, low-density and high-density lipoprotein cholesterol concentrations were 5.60 (0.77), 3.73 (0.70) and 1.26 (0.34) mmol/L after the A1 diet and 5.63 (0.81), 3.75 (0.75) and 1.27 (0.37) mmol/L after the A2 diets. We found no evidence that dairy products containing beta-casein A1 or A2 exerted differential effects (P > 0.05) on plasma cholesterol concentrations in humans.