Neuroprotective drugs to protect the brain against cerebral ischemia and reperfusion (I/R) injury are urgently needed. Mammalian cell-produced recombinant human erythropoietin (rhuEPO^M) has been demonstrated to have excellent neuroprotective functions in preclinical studies, but its neuroprotective properties could not be consistently translated in clinical trials. The clinical failure of rhuEPO^M was thought to be mainly due to its erythropoietic activity-associated side effects. To exploit its tissue-protective property, various EPO derivatives with tissue-protective function only have been developed. Among them, asialo-rhuEPO, lacking terminal sialic acid residues, was shown to be neuroprotective but non-erythropoietic. Asialo-rhuEPO can be prepared by enzymatic removal of sialic acid residues from rhuEPO^M (asialo-rhuEPO^E) or by expressing human EPO gene in glycoengineered transgenic plants (asialo-rhuEPO^P). Both types of asialo-rhuEPO, like rhuEPO^M, displayed excellent neuroprotective effects by regulating multiple cellular pathways in cerebral I/R animal models. In this review, we describe the structure and properties of EPO and asialo-rhuEPO, summarize the progress on neuroprotective studies of asialo-rhuEPO and rhuEPO^M, discuss potential reasons for the clinical failure of rhuEPO^M with acute ischemic stroke patients, and advocate future studies needed to develop asialo-rhuEPO as a multimodal neuroprotectant for ischemic stroke treatment.