This report describes the results of our initial enzymological characterization of a homogeneous preparation of DNA polymerase alpha that we have purified from cultured human KB cells. Although the enzyme is most reactive with duplex DNA substrates that contain short gaps (optimally activated) in incubations that require Mg2+, the polymerase possesses the intrinsic capacity to copy the initiated ribohomopolymer template, (A)-n, (dT)-200, at low rates in the presence of Mn2+. Because of the preponderance of DNA polymerase alpha in actively multiplying vertebrate cells, it is probable that this low level of activity comprises the majority of the ribopolymer copying activity that can be detected in crude tissue extracts. The presence of contaminating or associated deoxyribonuclease activities can be excluded from the purified enzyme to levels of 10(-4) to 10(-7) of the polymerase activity. The mechanism of polymerization on activated DNA under optimum conditions is moderately processive, with 11 +/- 5 nucleotides incorporated per polymerization cycle. The polymerase is unable to work at nicks or at short gaps of approximately 20 to 30 nucleotides in length, and it measures a surprisingly invariant effective template length on optimally activated DNA and on DNA molecules that have been gapped to varying extents with Escherichia coli exonuclease III. In the "Appendix" we present an amplification of the theoretical formulation of Bambara et al. (Bambara, R. A., Uyemura, D., and Choi, T. (1978) J. Biol. Chem. 253, 413--423) that permits the use of DNA polymerases with significant associated 3' leads to 5'-exonuclease activities for the accurate measurement of average template lengths (gap sizes) and titration of usable 3'-hydroxyl primer termini in gapped, duplex DNA substrates.