Retest reliability for plasma cholesterol and triglyceride is important diagnostically and in the planning of studies. Coefficients of variation of about 8% for cholesterol and 25% for triglyceride are reported here during Visits 1 and 2 of the Lipid Research Clinics Prevalence Study in 1972-1975. Retest reliability showed little or no dependence on other factors studied. It would not be unusual to observe apparent cholesterol changes of 50 mg/dl and occasionally 100 mg/dl with no obvious explanation. Apparent random triglyceride changes may be as high as several hundred mg/dl. It is of clinical importance that two thirds of persons in the upper quintile of cholesterol and/or triglyceride at first measurement remain in this rank and one third rank lower on retest. Multiple observations of these lipid values are recommended both for clinical use and for establishment of study baselines.