Anemia is common in hospital patients, being found in about half of the automated complete blood counts (CBCs) done on adults aged 20+ at our hospital. Often the reason for the anemia is immediately apparent (post-operative state, end stage renal disease without erythropoietin treatment), but often it is not. Many tests are available for the differential diagnosis of the causes of the anemia. Morphologic evaluation of a Wright-stained blood smear by a skilled observer is labor intensive, but sometimes useful in the differential diagnosis of anemia. When unexplained anemia is identified in a patient on the Medical Service at our teaching hospital, the ward team may also send an intern or a medical student to check out the smear. The potential diagnostic yield of any of these evaluations (the skilled observer, the intern, the student) is unknown. We did a prospective evaluation of 202 consecutive adults with initially unexplained anemia (Hb <12 in men, <11 in women). Using accepted, pre-established criteria for etiologic diagnosis of the causes of anemia, and available data (plus additional smear evaluations by ourselves, ferritins, free erythrocyte protoporphryns, and a few other tests) we felt we were able to establish the causes of anemia in 86% of the patients. We also referred to standard morphologic criteria for diagnosing specific blood disorders in order to see how much could be learned from the blood smears of these patients, alone. Of 147 patients whose blood smears were of acceptable quality, 30 (21%) of the blood smears were diagnostic (or close to diagnostic), and an additional 46% were supportive of the correct diagnosis. For example, iron deficiency anemia was considered the morphologic diagnosis if the blood smear showed severe hypochromia, “pencil-form” elliptocytes, and thrombocytosis. Hypochromia alone would be considered supportive of the diagnosis of iron deficiency, but require more consideration of alternatives (thalassemia trait, chronic disease, etc). Thus, in evaluating anemia in adults, a good peripheral blood smear, carefully evaluated by a trained observer, is likely to be diagnostic, or very helpful, in 20% or more of the patients.