The objective was to study whether a yoghurt containing isolated soya protein with standardised levels of isoflavones, cotyledon soya fibres and soya phospholipids is more effective in lowering total and LDL-cholesterol than a placebo. One hundred and forty-three subjects were randomised to the soya group (n 69) or to the placebo (n 74). The mean baseline levels were 7·6 and 5·1mmol/l for total and LDL-cholesterol, respectively. Fasting serum lipoproteins were assessed five times during the 8-week intervention period, and 4 weeks thereafter. The results were analysed by a mixed model for unbalanced repeated measurements. During the intervention, there were highly significant differences in lipid-lowering effect in favour of the active soya intervention group compared with the control group. The significant differences were for total cholesterol (estimated mean difference 0·40mmol/l; P<0·001), LDL-cholesterol (0·39mmol/l; P<0·001), non-HDL-cholesterol (0·40mmol/l; P<0·001) and for the total:HDL-cholesterol ratio (0·23; P=0·005). There was no difference in the effects on HDL-cholesterol, triacylglycerols or homocysteine. The lipid-lowering effect occurred within 1–2 weeks of intervention, and was not due to weight loss. The safety profile for active soya was similar to the placebo group, except for gastrointestinal symptoms, which caused a significantly higher dropout rate (fourteen v. three subjects) among the subjects taking active soya.