This report suggests modest changes in the criteria used for the diagnosis of ET and allows tentative recommendations concerning therapy. As outlined in Table I, we believe that absent stainable marrow iron does not necessarily indicate iron deficiency in these patients and that the serum ferritin and RBC mean corpuscular volume should be incorporated in this assessment. Normal values speak strongly against iron-deficient erythropoiesis. A search for the bcr/abl gene rearrangement should be included with the marrow karyotype to exclude CML. Finally, cytogenetic data and morphologic study of the marrow should be used to be certain that a MDS should not be considered. It may be that measurements of serum thrombopoietin levels may be useful in the future. Nonetheless, in principle, ET remains a diagnosis of exclusion as we have originally suggested. For therapy, HU remains an excellent choice for the older patient at risk for thrombosis. Nonetheless, no myelosuppressive therapy remains a perfectly viable option, particularly for the young patient and the older with low thrombotic risk. The roles of anagrelide and alpha interferon in this setting have not been fully defined. Experience with both has still been relatively short. It would be ideal if prospective, randomized trials could be mounted to address these questions. We conclude with confidence that return to older approaches such as 32P and AA in patients who fail on HU is to be discouraged. The use of anagrelide or interferon alfa seems to be a much more appropriate approach. We have not investigated the role of antithrombotic agents such as aspirin in ET. In PV, the combination of aspirin, 300 mg three times daily, and dipyridamole, 75 mg three times daily, failed to reduce the rate of thrombosis and was associated with an increased rate of hemorrhage. It is rational to suggest that lower doses of aspirin (ie, < 325 mg daily) might be associated with less hemorrhage and, perhaps, a beneficial effect on thrombosis. This remains to be shown.