Background: Due to the potential risk of iron supplementation in iron replete children, it is important to properly identify children who may require iron supplementation. However, assessment of the iron status has proven to be difficult, especially in children living in areas with high infection pressure (including malaria). Aims and Methods: Biochemical iron markers were compared to bone marrow iron findings in 381 Malawian children with severe anaemia. Results: Soluble transferrin receptor/log ferritin (TfR-F index), using a cut-off of 5.6, best predicted bone marrow iron stores deficiency (sensitivity 74%, specificity 73%, accuracy 73%). In order to improve the diagnostic accuracy of ferritin or sTfR as a stand-alone marker, the normal cut-off value needed to be increased by 810% and 83% respectively. Mean cell haemoglobin concentration (MCHC), using a cut-off of 32.1 g/dl, had a sensitivity of 67% and specificity of 64% for detecting iron stores deficiency. Conclusion: TfR-F index incorporated the high sensitivity of sTfR, a proxy for cellular iron need, and the high specificity of ferritin, a proxy for iron stores. In areas with a high infection pressure, the TfR-F index best predicted iron deficiency. However, in settings where diagnostic tests are limited, MCHC may be an acceptable alternative screening test.