The accuracy of nucleotide selection by wild type, L88 mutator, and CB120 antimutator T4 DNA polymerases has been compared by measuring both stable incorporation of complementary and noncomplementary nucleotides into polymer and the DNA-dependent conversion of deoxynucleoside triphosphate to monophosphate. The increased accuracy of the CB120 antimutator enzyme is shown by a ratio of utilization of incorrect to correct nucleotides with poly(dA)-poly(dT) as template which is only 10 to 30% of that of the wild type enzyme. In contrast, the ratio of incorrect to correct nucleotide utilized by the L88 mutator enzyme with this template was higher than that of the wild type enzyme, in agreement with the report of Hershfield (Hershfield, M.S. (1973) J. Biol, Chem. 248, 1417-1423). The antimutator, mutator, and wild type enzymes each have a much higher apparent Km for the noncomplementary nucleotides than for complementary nucleotides with this template. The L88 mutator polymerase has a higher "Km" for poly(dA)-poly(dT) than the wild type enzyme in reactions with both complementary and noncomplementary nucleotides. The antimutator polymerase has an elevated "Km" for polymer only for reactions with noncomplementary nucleotides. The wild type and L88 mutator enzymes also utilized both noncomplementary nucleotides much more frequently than the CB120 antimutator enzymes with poly [d(A-T)] as the template-primer. The T4 gene 32DNA unwinding protein appears to facilitate the correct reading of this template since it decreases the ratio of incorrect nucleotides utilized by both the wild type and L88 mutator polymerases.