The aim of this study is to evaluate the accuracy of capillary hemoglobin (Hb) measurements in detecting anemia among low-income toddlers (aged 12-35 months) and pregnant women. In analyses of data among toddlers from Kansas City (n = 402) and St. Louis, Missouri (n = 236), and pregnant women at <20 weeks gestation from Cleveland, Ohio (n = 397), we compared subjects' anemia status based on capillary Hb concentrations in finger puncture samples as measured by the HemoCue system with their anemia status based on venous Hb concentrations as measured by the HemoCue and Coulter Counter. The sensitivity of capillary blood analyses in identifying cases of anemia was 32.8% (95% Confidence Intervals (CI): 21.0%-46.3%), among Kansas City toddlers, 59.7% (95% CI: 45.8%-72.4%) among St. Louis toddlers, and 66.7% (95% CI: 46.0%-83.5%) among pregnant women in Cleveland; the corresponding specificities were 97.7%, 86.6%, and 96.7%, respectively. The correlation between HemoCue and Coulter Counter measurements of venous Hb (0.9) was higher than that between HemoCue measurements of capillary and venous blood (0.8). The results show that Hb measurements of capillary blood with HemoCue were not optimal for determining the anemia status of toddlers and pregnant women.