The presence of erythropoietin (Epo) receptors on cells besides red blood cell precursors, such as cancer cells or megakaryocyte precursors, can lead to side effects during Epo therapy including enhanced tumor growth and platelet production. It would be ideal if the action of Epo could be limited to erythroid precursors. To address this issue, we constructed single-chain variable fragment (scFv)-Epo fusion proteins that used the anti-glycophorin 10F7 scFv amino-terminal to Epo analogues that would have minimal activity alone. We introduced the Epo mutations N147A, R150A and R150E, which progressively weakened receptor affinity in the context of Epo alone, as defined by cell proliferation assays using TF-1 or UT-7 cells. Fusion of these mutant proteins to the 10F7 scFv significantly rescued the activity of the mutant proteins, but had a relatively small effect on wild-type Epo. For example, fusion to the 10F7 scFv enhanced the activity of Epo(R150A) by 10- to 27-fold, while a corresponding fusion to wild-type Epo enhanced its activity only up to 2.7-fold. When glycophorin was blocked by antibody competition or reduced by siRNA-mediated inhibition of expression, the activity of 10F7 scFv-Epo(R150A) was correspondingly reduced, while such inhibition had essentially no effect on the activity of 10F7 scFv-Epo(wild-type). In addition, potent stimulation of Epo receptors by 10F7 scFv-Epo(R150A) was observed in long-term proliferation and viability assays. Taken together, these results indicate that a combination of targeting and affinity modulation can be used to engineer forms of Epo with enhanced cell-type specificity.