Objective - To evaluate the long term efficacy of diets in lowering serum cholesterol concentration. Design - Descriptive overview of 16 published controlled trials of six months' duration or longer. Setting - Trials had been conducted in hospital clinics (6), industry (3), mental hospitals or institutions (3), and in general populations (4). Patients - Trials had been conducted in high risk subjects (5), in unselected healthy subjects (6), or for secondary prevention in patients with coronary heart disease (5). Women were included in only four trials. Interventions - Diets equivalent to the step 1 diet were employed in eight trials, with individual intervention by dietitians (3) or occupational physicians (2) or with population advice (3). Intensive diets which were more rigorous than the step 2 diet were employed in eight trials. Main outcome measures - Net change in serum total cholesterol concentration in subjects receiving treatment with diet compared with values in control subjects after six months to 10 years. Results - In five trials with the step 1 diet as individual intervention the net reduction in serum cholesterol concentration ranged from 0% to 4.0% over six months to six years. In trials with population education reductions in cholesterol concentrations were 0.6-2.0% over five to 10 years. When population and individual dietary advice were combined changes in cholesterol concentration ranged from a fall of 2.1% to a rise of 1.0% over four to 10 years. Diets more intensive than the step 2 diet reduced serum cholesterol concentration by 13% over five years in selected high risk men in the population; by 6.5-15.1% over two to five years in hospital outpatients; and by 12.8-15.5% over one to four and a half years in patients in institutions. Conclusions - The response to a step 1 diet is too small to have any value in the clinical management of adults with serum cholesterol concentrations above 6.5 mmol/l. Current guidelines recommend screening of serum cholesterol concentration in healthy subjects, followed by treatment with a step 1 diet. The guidelines should be reviewed to provide a more realistic estimate of the effect of a step 1 diet and of the likely need for lipid lowering drugs.