The reliability of the assay of endogenous erythroid colony (EEC) formation in serum-free, cytokine-free collagen-based media was investigated in a multicentric study including 140 patients with polyglobuly (80 polycythemia vera (PV), 54 secondary erythrocytosis (SE), six idiopathic erythrocytosis (IE)) and 10 healthy donors. In each center, EEC assays were performed in parallel with progenitor cells from bone marrow (BM) and peripheral blood (PB); two commercialized media and 'low' and 'high' cell plating densities were tested. Negativity of EEC assays was considered certain only when sufficient BFU-E growth was obtained in control cultures with cytokines. In the two media, EEC formation was specific - never observed in cultures of healthy donors or SE patients - and comparable. BM EEC assays were positive (presence of eythroid colonies) for 75% ('low' plating) to 100% ('high' plating) of PV patients; PB EEC assays were positive for 83.3% ('low' plating) to 93.7% ('high' plating) of PV patients (differences not significant). Depending on the medium, 86.2-93.7% of patients with a positive BM EEC assay had a positive PB EEC assay. Hence, a standardized collagen-based EEC assay can be performed with either BM or PB progenitors; the EEC assay described here is positive for at least 75% of PV patients when a single EEC assay is performed, and for at least 94% of PV patients when both BM and PB EEC assays are performed.