The erythropoietic effects of lenalidomide are cytokine dependent, suggesting that the erythroid hematologic improvement (HI-E) rate may be augmented by combined treatment (CT) with recombinant human erythropoietin (rhu-EPO) in myelodysplastic syndrome (MDS). In the present study, we explored the benefits of CT and the relationship between lenalidomide pharmacokinetics and hematologic toxicity in transfusion-dependent patients with low- to intermediate-1-risk MDS who failed prior rhu-EPO. In stage I, patients received 10 or 15 mg/d of lenalidomide monotherapy. At week 16, erythroid nonresponders (NRs) were eligible for CT with rhu-EPO 40 000 U/wk. Among 39 patients, HI-E response rate to monotherapy was 86% (6 of 7) in del(5q) and 25% (8 of 32) in non-del(5q) patients (10 mg, 17.7%; 15 mg, 33.3%). Twenty-three patients proceeded to CT, with 6 (26.0%) achieving HI-E. In 19 non-del(5q) patients, 4 (21.1%) showed HI-E. Mean baseline serum EPO in non-del(5q) patients was lower in monotherapy and CT responders than in NR (not statistically significant). Thrombocytopenia was significantly correlated with lenalidomide area under the plasma concentration-time curve (P = .0015), but severity of myelosuppression did not. The benefits of lenalidomide plus rhu-EPO are currently under investigation in a phase 3 Eastern Cooperative Oncology Group (ECOG)-sponsored intergroup study. This study is registered at www.clinicaltrials.gov as NCT00910858.