The genomic and complementary DNAs of the human erythropoietin receptor (hEpo-R) have been isolated and characterized from a genomic placental library and from two cDNA libraries prepared from bone marrow and fetal liver. The five different partial cDNAs isolated were aberrant in the predicted reading frames as compared with the Epo-R protein sequence, because all retained insert sequences that may represent splicing intermediates (three clones), cloning artifact (one clone), or a new sequence at a splice junction (one clone) of the gene. The cDNAs were used to isolate several genomic clones encompassing the complete hEpo-R gene. This gene, which encodes a 508-amino acid polypeptide chain of predicted M(r) 55,000, is organized into eight exons spread over 6 kb of DNA and exhibited a high degree of sequence homology (81.6% in the coding region) and structural organization with its murine counterpart. Primer extension analysis indicated that the transcription initiation site is located 141 bp upstream of the initiation codon. Sequence homology 320 bp upstream of the cap site was significantly lower (60%) and diverged completely further upstream as compared with the murine gene. Similarly, the human and murine sequences were largely divergent downstream of the stop codon, indicating that a strong conservation during evolution was restricted to the coding sequence of the Epo-R protein. The 320-bp region upstream of the cap site does not contain the typical TATA or CAAT boxes present in many tissue-specific genes, but does include potential binding sites for the ubiquitous Sp1 and the erythroid-specific GATA-1 trans-activating factors. These boxes are well conserved in sequence and position relative to the cap site within the promoter region of the human and murine genes, but the CACCC boxes present in the murine gene are absent in the human gene.