Several properties of the viral RNA-dependent DNA polymerases and of rabbit globin mRNA make it possible to consider synthesis of the globin gene in vitro. These enzymes copy an RNA template using a short sequence of complementary nucleotides as a primer. Furthermore, globin mRNA has a 3'-terminal sequence of adenylic acid residues that make it particularly suitable as a template, since oligo(dT) can be annealed to a specific site on the mRNA. This small primer could phase the DNA polymerase, possibly ensuring that replication is initiated from that end of the globin message. We have used this approach and find that purified mRNA is an efficient template for the polymerase enzyme. The reaction requires the RNA template and the four deoxyribonucleoside triphosphates, and it is markedly stimulated by the addition of oligo(dT). Consistent with the expectation that the oligo(dT) uniquely phases the polymerase at an adenine-rich region in the globin message, oligo(dG), oligo(dC), and oligo(dA) fail to serve as primers. The product has a density intermediate between that of DNA and RNA, and shifts to a lighter DNA density after treatment with base. Further, it is specifically complementary to globin mRNA and sediments slightly faster in an alkaline sucrose gradient than a DNA standard that has a molecular weight of 129,000. The data suggest that a major portion of the DNA product is a sequence of at least 500 bases, about 50 more than would be necessary to encode rabbit globin. The potential usefulness of this interesting product is discussed.