Taq and T7 DNA polymerases have become basic molecular biology "tools" for DNA sequence analysis. However, Taq, unlike T7 DNA polymerase, is strongly biased against the incorporation of 2',3'-dideoxynucleotide triphosphates (ddNTPs) indicating very different substrate selectivities. Equilibrium binding and rate constants were measured for 2',3'-ddNTPs as well as for several other 3'-substituted terminators and compared to 2'-deoxynucleotide substrates (dNTPs). In steady-state experiments, Taq Pol I was strongly biased in favor of dATP1 over ddATP incorporation by about 700 to 1, in contrast to T7 DNA polymerase which showed a preference of only about 4 to 1. Manganese reduced but did not eliminate selectivity against 2',3'-ddNTPs. Transient kinetic traces indicated different rate-limiting steps for substrate and terminator incorporation. Further mechanistic studies showed that the binding constants for substrates and terminators were equivalent. However, the rate constants for phosphodiester bond formation for 2',3'-ddNTPs were 200-3000-fold lower than for dNTPs. Alternative terminators showed only slight improvements. The data were consistent with a model in which both substrates and terminators undergo ground-state binding followed by formation of a tight-binding Enz.DNA.Nucleotide complex. Immediately after complex formation, substrates undergo a rapid nucleoside phosphoryl transfer reaction. However, the reaction rates for terminators were slower presumably due to misalignment of reactive groups in the active site. Thus, the strong bias that Taq DNA polymerase shows against terminators is due to a very slow "chemistry" step. Such a strong bias has several kinetic consequences for DNA sequence patterns. These consequences are discussed in the text.