Blood was sampled from nine healthy subjects weekly for 10 weeks, under controlled conditions. Eighteen serum constituents were measured in duplicate in each sample on a single occasion. In this way analytic variation was minimized. Considerable variations in mean values and standard deviations of results were observed among the subjects. Physiological control of serum calcium and magnesium concentrations appeared to be more precise than duplicate determinations in the laboratory. Compared with a previous study from this laboratory, significantly different estimates of personal variation were noted for sodium, chloride, urea nitrogen, cholesterol, and LDH in a group matched by age, race, and sex. At the same time, the present study group, homogenous with respect to these demographic factors (and also occupation) showed as great a diversity in the mean concentrations of most constituents as was seen earlier in a larger, more heterogenous group of normal individuals. These findings indicate that certain blood parameters depend more on personal characteristics than on broad demographic factors.