Erythropoietin (EPO) mediates a wide range of neuroprotective activities, including amelioration of disease and neuroinflammation in rat models of EAE. However, optimum dosing parameters are currently unknown. In the present study, we used a chronic EAE model induced in mice by immunization with the myelin oligodendrocyte glycoprotein peptide (MOG35-55) to compare the effect of EPO given with different treatment schedules. EPO was administered intraperitoneally at 0.5, 5.0 or 50 microg/kg three times weekly starting from day 3 after immunization (preventive schedule), at the onset of clinical disease (therapeutic schedule) or 15 days after the onset of symptoms (late therapeutic schedule). The results show that EPO is effective even when given after the appearance of clinical signs of EAE, but with a reduced efficacy compared to the preventative schedule. To determine whether this effect requires the homodimeric EPO receptor (EPOR2)-mediated hematopoietic effect of EPO, we studied the effect of carbamylated EPO (CEPO) that does not bind EPOR2. CEPO, ameliorated EAE without changing the hemoglobin concentration. Another non-erythropoietic derivative, asialoEPO was also effective. Both EPO and CEPO equivalently decreased the EAE-associated production of TNF-alpha, IL-1beta and IL-1Ra in the spinal cord, and IFN-gamma by peripheral lymphocytes, indicating that their action involves targeting neuroinflammation. The lowest dosage tested appeared fully effective. The possibility to dissociate the anti-neuroinflammatory action of EPO from its hematopoietic action, which may cause undesired side effects in non-anemic patients, present new avenues to the therapy of multiple sclerosis.