Serum ferritin (SF) values </=10 micro g/l are diagnostic of absent Bone Marrow Iron (BMI) stores and therefore of iron deficiency (ID). However, SF, which may be elevated as a part of acute phase reaction, is an unreliable indicator of BMI stores in the setting of chronic disorders, making it difficult to diagnose ID in these patients. Thus, in chronic disorders (CD) such as tuberculosis, bone marrow examination is the only reliable way to establish ID. This study was done in order to identify levels of SF that would be indicative of absent BMI stores and also to study a combination of hematological and biochemical parameters that would be helpful in raising the predictive power of SF in patients of tuberculosis. Fifty-five tuberculosis patients were studied and classified into Iron Deplete (ID) and Iron Replete (IR) based on BMI. Raising the cut-off values of SF from </=10 micro g/l to </=30 micro g/l diagnosed 88% of ID cases correctly, as compared with 61% when cut-off levels of </=10 micro g/l were used. At cut-off values higher than 30 micro g/l, the sensitivity was markedly reduced. Therefore, raising cut-off levels of SF to </=30 micro g/l was most effective in predicting absent BMI, especially in a population where ID is highly prevalent. Combination of SF </=30 micro g/l with mean corpuscular volume (MCV), erythrocyte sedimentation rate (ESR) and total iron binding capacity (TIBC) did not improve the predictive power of SF further. Also, 89.5% cases could be correctly classified by logistic regression equations using SF with ESR and C- reactive protein (CRP).