Much epidemiologic research is based on estimation of an association between a putative risk factor and a health outcome--for example, plasma concentration of lipoproteins and ischemic heart disease. Since the repeatability of a risk factor measurement determines, in part, the ability to ascertain its association in populations, the Atherosclerosis Risk in Communities (ARIC) Intraindividual Variability Study was conducted to estimate various components of variation in analyte data and to estimate the repeatability of these measurements. A total of 40 subjects (17 males and 23 females) from Forsyth County, North Carolina, Minneapolis, Minnesota, Jackson, Mississippi, and Washington County, Maryland, were studied in 1988. Fasting blood was collected three times from each subject, with a 1- to 2-week interval between each visit. The contributions of between-person variability, within-person variability, and processing and assay variability were estimated. From these components, the reliability coefficient, R, the correlation between measures made at repeat visits, was estimated. R was above 0.85 for total cholesterol, high density lipoprotein cholesterol, low density lipoprotein cholesterol, triglycerides, and lipoprotein(a). Low repeatability was obtained for apolipoprotein A-I (R = 0.60). High density lipoprotein subfractions 2 and 3 were intermediate in repeatability. Reliability coefficients from the ARIC Intraindividual Variability Study are generally higher than those found in other studies, and this is related to relative variability in populations studied, to the time between measurements, and to differences in laboratory variability. Only for apolipoprotein A-I would the findings strongly suggest the need to adjust for measurement variability in estimation using one of these analytes as an independent variable.