In elderly patients the diagnosis of iron deficiency from full blood count indices is often difficult. We assessed an automated technique (numerical data of the erythrogram; Technicon H*l) by which the proportions of microcytic ( < 60 fl) and/or hypochromic ( < 28 g·dl-1) red blood cells are determined. Of 472 elderly patients investigated, 100 (21%) were found to have iron deficiency (plasma ferritin ≤45 ng·ml-1). Less than two-thirds of patients with iron-deficient erythropoiesis (anaemia or microcytosis) had increased proportions of hypochromic and/or microcytic red blood cells. Furthermore, the erythrogram was not sensitive in detecting latent or early iron deficiency. The erythrogram also lacked specificity for iron deficiency anaemia as many patients with mild normocytic anaemia associated with chronic inflammatory disease had increased proportions of hypochromic and/or microcytic red blood cells. Although patients with iron deficiency had increased proportions of hypochromic normocytic (p < 0.01) and normochromic microcytic red blood cells (p < 0.05) compared to those with chronic inflammatory disease and normal or raised iron stores (ferritin ≥100 ng·ml-1, n = 32). there was a large overlap between these two groups, and the grossly elevated erythrogram results in patients with iron deficiency were almost always associated with a mean cell volume (MCV) < 80 fl, whereas none of the patients with chronic inflammatory disease and normal or raised iron stores had an MCV < 80 fl. Thus the erythrogram does not appear to be of value in the routine assessment of iron status in elderly patients.