Plasma cholesterol and triglyceride levels were determined, on each of two AutoAnalyzer systems in 11 healthy subjects, weekly over a 10-week and monthly over a 12-month period. Analytical variation was 1-2% for cholesterol and 2-5% for triglyceride. Cholesterol and triglyceride values on frozen quality control serum pools were not indicative of absolute values on fresh plasma. Even though the two AutoAnalyzer systems averaged within 1-2 mg/dl for triglyceride and cholesterol on the serum quality control pools during the 12-month period, the two systems differed by 7-8 mg/dl on fresh or frozen plasma samples. The coefficient of physiological variation on the 10 weekly samples averaged 5% (range 3-10%) for plasma cholesterol and 18% (range 9-27%) for plasma triglyceride. Analysis of the monthly samples suggested significant (P less than 0.05) seasonal trends: cholesterol was highest in the winter months and lowest in October, whereas triglyceride was highest in January and February and lowest in May and December. We conclude that intra-individual variation can be an important source of error in attempting to make a genetic diagnosis of hyperlipidemia and/or in evaluating hypolipidemic regimens in a given subject.