Objective: To assess anemia incidence and persistence in low-income preschool children in the United States. Study design: Using 2000 to 2004 data from Center for Disease Control and Prevention's Pediatric Nutrition Surveillance System, we constructed 4 cohorts. Children in each cohort had a baseline hemoglobin measurement at either age 12 ± 2 months (n = 583,149), 18 ± 2 months (n = 399,223), 24 ± 2 months (n = 382,605), or 36 ± 2 months (n = 300,817) and a follow-up hemoglobin measurement 12 ± 2 months later, when they were approximately 24, 30, 36, or 48 months old. Defining anemia as a hemoglobin level <11.0 g/dL (<24 months old) or hemoglobin <11.1 g/dL (≥24 mo), we calculated anemia incidence and persistence in each cohort and used multiple logistic regression to identify associated factors (race, sex, birthweight, height, weight, breastfeeding). Results: Anemia incidence declined with age. Persistence remained approximately 30%. In each cohort, 70% of follow-up anemia cases were incident. Compared with white children, black children had greater odds of incident anemia at each follow-up age (odds ratio [OR], 1.84-2.09), while Native American children had lower odds at 36 and 48 months of age (OR, 0.68, 0.65). Both Asian and black children had greater odds of persistent anemia than white children at each age (OR, 1.73-2.60). Conclusions: Most follow-up anemia in each cohort was incident, underscoring the importance of anemia prevention throughout early childhood in this population. Investigation of the causes of anemia is warranted.