Objective - To determine the extent to which reducing the saturated fatty acid composition of a ubiquitously used cooking oil influenced changes in cholesterol concentration in the population during a five year intervention programme in Mauritius. Design - Cross sectional surveys in 1987 and 1992 determined mean total cholesterol concentrations in the population. A random sample of respondents in the 1992 survey completed a nutrition questionnaire that included questions on diet in the previous 24 hours. Setting - Mauritius. Intervention - In 1987 the government of Mauritius changed the composition of the commonly used cooking oil from being mostly palm oil (high in saturated fatty acids) to being wholly soya bean oil (high in unsaturated fatty acids). Subjects - 5080 and 5162 subjects in 1987 and 1992 cross sectional surveys. 2059 subjects aged 30-64 years were randomly selected from the respondents of the 1992 survey to take part in the nutrition survey. Main outcome measures - Fatty acid composition of phospholipids in pooled serum samples from men and women from the two surveys; measured and predicted change in serum cholesterol concentration. Results - From 1987 to 1992 total cholesterol concentrations fell significantly by 0.79 mmol/l (P < 0.001) in men and 0.82 mmol/l (P < 0.001) in women. The estimated intake of saturated fatty acids decreased by 3.5% of energy intake in men and by 3.6% in women, and the intake of polyunsaturated fatty acids increased by 5.5% and 5.6% of energy intake, respectively. These changes were reflected in changes in the far, acid composition of serum phospholipids, and according to Keys' formula these changes explained much of the decrease in serum cholesterol concentrations (predicted decrease of 0.38 mmol/l in men and by 0.40 mmol/l in women). Conclusion - Dietary changes that entailed a reduction in the saturated fat content of a ubiquitous cooking oil explained most of the observed decrease in serum cholesterol concentration over five years in the population of Mauritius.