Epidemiologists usually employ measures of variability of physiological traits such as blood pressure and cholesterol only to determine confidence intervals or statistical significance. For evolutionary biologists population variability per se has proven of interest. This article explores the applicability of this perspective to the analysis of human physiology, using data from the Framingham Heart Disease Study and the National Health and Nutrition Examination Surveys. The nonrandom patterns of variability observed suggest that examination of the degree and pattern of heterogeneity within a population may provide information not evident from the analysis of mean values.