Despite decades of intensive research, lead (Pb) toxicity still remains one of the most frequently investigated subjects in environmental health. Whole blood lead (BPb) is usually used to evaluate Pb exposure for both screening and clinical diagnosis. However, it is generally recognized that BPb is not a sensitive biomarker for Pb exposure in hematological studies. Considering hematocrit (HCT) variation in different situations, HCT-adjusted BPb or erythrocyte Pb (EPb) may be more relevant when evaluating the hematotoxicity of blood Pb. Data collected from 855 preschool children, 3-7 years of age, allowed us to examine the relationship between EPb and hemoglobin (Hb) levels. Multivariate linear regression was performed to determine the significance of EPb as predictor of Hb after covariate adjustment; then, mean differences of Hb levels between quartiles of EPb and BPb (1st quartile as reference) were determined using ANOVA followed by Student's t test. The dose-response curve between EPb and HCT was plotted using locally weighted scatterplot smoothing (LOWESS) method. A doubling of EPb was associated with a 2.44 g/L decrease in Hb level. Compared to the 1st quartile group of EPb, the 3rd and 4th quartile groups showed significant decreases in Hb levels (3.01 and 3.97 g/L, respectively). Compared to the 1st quartile group of BPb, the 2nd quartile group showed a decrease in Hb levels (0.63 g/L), while the 3rd and 4th quartile groups showed increases in Hb levels (0.78 and 1.45 g/L, respectively). Increased EPb levels are significantly associated with decreased Hb levels in preschool children. HCT must be taken into consideration in investigating the hematological effects of Pb. Compared to BPb, EPb or HCT-adjusted BPb appear as a more effective biomarker to interpret the hematotoxicity of lead. Furthermore, blood erythrocytes are not only a repository of Pb but also a primary target of its toxicity.