Due to limited penetration of the BBB, many therapeutic agents in clinical use require higher doses in order to reach effective concentrations in brain. In some instances, these high doses elicit severe side effects. In the case of erythropoietin (EPO), an established neuroprotectant against ischemic brain injury, its low BBB permeability requires such a high therapeutic dose that it can induce dangerous complications such as polycythmia and secondary stroke. The purpose of this study is to generate a modified EPO that has increased facility crossing the BBB without losing its neuroprotective element. We have engineered a fusion protein (EPO-TAT) by tagging a protein transduction domain derived from HIV TAT to the EPO protein. This sequence enhanced the capacity of EPO to cross the BBB in animals at least twofold when IP administered and up to five-fold when IV administered. In vitro experiments showed that this EPO fusion protein retained all its protective properties against neuronal death elicited by oxygen-glucose deprivation and NMDA insults. The needed therapeutic dose of the EPO-TAT was decreased by ~10-fold compared to that of regular EPO to achieve equivalent neuroprotection in terms of reducing volume of infarction induced by middle cerebral artery occlusion in mice. Our results support the approach of using a protein transduction domain coupled to therapeutic agents. In this way, not only can the therapeutic doses be lowered, but agents without BBB permeability may now be available for clinical applications.