Hb and the per cent transferrin saturation (TS) distributions of 742 black and 3074 white women drawn from a large probability sample of the USA civilian noninstitutionalized population (HANES I) were examined to ascertain the magnitude and nature of an observed systematic difference in Hb levels. This significant Hb difference was primarily related to a different relationship between Hb and TS for blacks and white, and not related to differences in TS between the races. This difference between the races in the relationship between Hb and TS explained almost all of the difference in Hb concentration at the mean, the median and higher percentiles, and half or more of the difference at the 1st and 3rd percentiles of the cumulative Hb distributions of the white and black populations sampled. The practical implication of these findings is that white Hb diagnostic criteria (standards) are inappropriate for blacks, and vice versa, both at the mean and at the lower Hb levels of clinical interest. [This study is relevant to Fe deficiency anemia epidemiology.].