DNA polymerases are complex enzymes which bind primer−template DNA and subsequently either extend or excise the terminal nucleotide on the primer strand. In this study, a stopped-flow fluorescence anisotropy binding assay is combined with real-time measurements of a fluorescent adenine analogue (2-aminopurine) located at the 3‘-primer terminus. Using this combined approach, the exact time course associated with protein binding, primer terminus unstacking, and base excision by the 3‘ → 5‘ exonuclease of bacteriophage T4 (T4 pol) was examined. T4 pol binding and dissociation kinetics were found to obey simple kinetics, with identical on rates (kon = 4.6 × 108 M-1 s-1) and off rates (koff = 9.3 s-1) for both single-stranded primers and double-stranded primer−templates (at 100 μM Mg2+). Although the time course for T4 pol−DNA association and dissociation obeyed simple kinetics, at suboptimal Mg2+ concentrations (e.g., 100 μM), non-first-order sigmoidal kinetics were observed for the base-unstacking reaction of the primer terminus in double-stranded primer−templates. The observed sigmoidal kinetics for base unstacking demonstrate that T4 pol is a hysteretic enzyme [Frieden, C. (1970) J. Biol. Chem. 245, 5788−5799] and must exist in two DNA bound conformations which differ greatly in base-unstacking properties. A Mg2+-dependent time lag of 10 ms is observed between primer−template binding and the beginning of the unstacking transition, which is 50% complete at 22 ± 1 ms after addition of 100 μM Mg2+. Following the hysteretic lag, a simple first-order primer terminus unstacking rate of 130 s-1 is resolved, which is protein and Mg2+ concentration-independent. For the processing of single-stranded primers, all kinetic complexity is lost, and T4 pol binding and primer end base-unstacking kinetics can be superimposed. These data reveal that the kinetic processing of double-stranded primer−template DNA by T4 pol is much more complex than that of single-stranded primers, and suggest that the intrinsic “switching rate” between the polymerase and exonuclease sites may be much faster than previously proposed.