The authors evaluated the utility of selective screening criteria for postpartum anemia developed by the Centers for Disease Control and Prevention (CDC) versus criteria developed among low-income women using prevalence-based screening principles. Pregnant women in Raleigh, North Carolina, were followed up to the postpartum visit in 1997-1999 (n = 345). Prevalence of postpartum anemia was 19.1%. Independent risk markers, arrived at through multivariate logistic regression, were multiparity (odds ratio (OR) = 1.5, 95% confidence interval (CI): 0.8, 2.9), obesity (OR = 3.0, 95% CI: 1.6, 5.5), anemia at 24-29 weeks' gestation (OR = 2.3, 95% CI: 1.2, 4.4), anemia before delivery (OR = 3.4, 95% CI: 1.8, 6.7), and not exclusively breastfeeding (OR = 2.8, 95% CI: 1.0, 7.7). Risk scores were calculated by counting risk markers present. Likelihood ratios were determined for all possible risk scores of our algorithm and CDC's algorithm. Anemia screening decisions differed depending on clinic anemia prevalence. For example, if low test thresholds are assumed, when clinic prevalence is 10%, women with risk scores >3 on the authors' algorithm and >0 on CDC's algorithm should be screened. The authors' algorithm, in combination with prevalence information, can save clinics more money than CDC's current algorithm because a broader range of likelihood ratios was obtained, indicating a better ability to distinguish high- from low-risk women. However, if resources are available, universal screening should be considered in high-prevalence settings.