Addressing the complex, multi-factorial causes of childhood anaemia is best done through integrated packages of interventions. We hypothesized that due to reduced child vulnerability, a “buffering” of risk associated with known causes of anaemia would be observed among children living in areas benefiting from a community-based health and nutrition program intervention. Cross-sectional data on the nutrition and health status of children 24–59 mo (N = 2405) were obtained in 2000 and 2004 from program evaluation surveys in Ghana, Malawi and Tanzania. Linear regression models estimated the association between haemoglobin and immediate, underlying and basic causes of child anaemia and variation in this association between years. Lower haemoglobin levels were observed in children assessed in 2000 compared to 2004 (difference -3.30 g/L), children from Tanzania (-9.15 g/L) and Malawi (-2.96 g/L) compared to Ghana, and the youngest (24–35 mo) compared to oldest age group (48–59 mo; -5.43 g/L). Children who were stunted, malaria positive and recently ill also had lower haemoglobin, independent of age, sex and other underlying and basic causes of anaemia. Despite ongoing morbidity, risk of lower haemoglobin decreased for children with malaria and recent illness, suggesting decreased vulnerability to their anaemia-producing effects. Stunting remained an independent and unbuffered risk factor. Reducing chronic undernutrition is required in order to further reduce child vulnerability and ensure maximum impact of anaemia control programs. Buffering the impact of child morbidity on haemoglobin levels, including malaria, may be achieved in certain settings.