A polyacrylamide gel assay is used to measure the kinetics of adding a single deoxyribonucleotide onto either a correctly matched or mismatched primer 3' terminus (on M13 template) for all possible DNA base pairs and mispairs using Drosophila melanogaster DNA polymerase alpha (Pol alpha) and avian myeloblastosis virus reverse transcriptase. The reverse transcriptase catalyzes chain extension from transition mispairs (Pur.Pyr and Pyr.Pur, where Pur is purine and Pyr is pyrimidine) more efficiently than polymerase alpha. Reverse transcriptase extends G(primer).T almost 20% as efficiently as it extends A.T, while Pol alpha's G.T extension efficiency is less than 1%. For transversion mispairs (Pur.Pur and Pyr.Pyr), reverse transcriptase extends C.T and T.T with greater efficiency than polymerase alpha, while polymerase alpha is more efficient at extending A.G and G.G mispairs. Reverse transcriptase and polymerase alpha extend the G.G mispair at an efficiency of only 10(-6) and 10(-5), respectively, compared with G.C extension. The extension data for the two polymerases are compared with previously reported nucleotide misinsertion data for the same enzymes (Mendelman, L. V., Boosalis, M. S., Petruska, J., and Goodman, M. F. (1989) J. Biol. Chem. 264, 14415-14423). While the results obtained with reverse transcriptase and Pol alpha differ in detail, some general rules are indicated: (a) Pur.Pyr and Pyr.Pur mispairs, especially G.T and T.G, are easy to insert and even easier to extend; (b) Pyr.Pyr mispairs, especially C.C, are difficult to insert and slightly easier to extend; (c) Pur.Pur mispairs, notably G.G, are harder to extend than to insert. The comparison also shows that reverse transcriptase extends almost all mismatches more efficiently than it forms them, G.G being the only mismatch having a significantly lower efficiency of extension than insertion. Polymerase alpha inserts A.A mismatches most efficiently, but extends them inefficiently, thereby reducing the probability that such transversion mutations will occur in vivo. We show theoretically that when mispaired primers compete with properly matched primers for extension by polymerase, the relative velocities of extension depend on the concentration of the next correct dNTP substrate. The extension velocities depart from Michaelis-Menten kinetics by exhibiting positive cooperativity with respect to substrate concentration.