BACKGROUND AND OBJECTIVE: It has been suggested that the determination of serum erythropoietin (sEpo) may be useful in distinguishing between polycythemia vera (PV), relative polycythemia and secondary polycythemia (SP), but no conclusive evidence has yet been provided for this. In the present work, we evaluated the role of sEpo in the differential diagnosis of polycythemia vera and its usefulness in the follow-up of PV patients. METHODS: sEpo was assessed in 190 patients with polycythemia of different etiologies. A follow-up study was carried out in some of these patients (27 with secondary polycythemia and 17 with polycythemia vera). RESULTS: sEpo levels were higher in SP than in PV and relative polycythemia. There were, however, differences with regard to the various etiologies of SP. Polycythemia related to congenital heart disorders showed the highest levels of sEpo of the SP. When a study was conducted, sEpo alone as a diagnostic parameter displayed an efficiency of more than 90% and the most discriminating value was 5 U/L. Using lower levels (below 2 U/L) and higher levels (above 12 U/L), it was possible to distinguish between SP and PV, although an important overlap was detected between these limits (approximately 50% of cases). The follow-up study showed that in half the patients with SP the levels of sEpo were at times < 12 U/L and at other times greater than this value. At least three determinations were necessary to detect an elevated reading. In PV after venesection there was an increase in sEpo in some cases, although most of the time there was no change. INTERPRETATION AND CONCLUSIONS: Using sEpo, it was possible to differentiate between PV and SP, despite an important overlap. A follow-up study demonstrated that the increase in sEpo was intermittent in SP and that in many of these cases more than one determination could be helpful.