We examined the relationship of serum ferritin to bone marrow iron stores in 73 anemic male medical inpatients with liver disease, alcoholism, chronic inflammatory disease, and malignancies. A correlation of r = 0.75 (P less than .00005) was found between serum ferritin and bone marrow iron stores (BMIS) for the entire group. Liver disease as manifested clinically or by increased levels of serum glutamic-oxaloacetic transaminase did not appear to significantly affect this relationship. Patients with folic acid deficiency did tend to have a disproportionate increase in ferritin in relation to BMIS, but this did not seem to destroy the usefulness of ferritin levels. A useful clinical rule seems to be that serum ferritin of greater than 100 ng/ml tends to exclude iron deficiency, and a level of less than 30 ng/ml tends to confirm decreased iron stores.