Quenching of redox active, intercalating dyes by guanine bases in DNA can occur on a femtosecond time scale both in DNA and in nucleotide complexes. Notwithstanding the ultrafast rate coefficients, we find that a classical, nonadiabatic Marcus model for electron transfer explains the experimental observations, which allows us to estimate the electronic coupling (330 cm(-1)) and reorganization (8070 cm(-1)) energies involved for thionine-[poly(dG-dC)](2) complexes. Making the simplifying assumption that other charged, pi-stacked DNA intercalators also have approximately these same values, the electron-transfer rate coefficients as a function of the driving force, AG, are derived for similar molecules. The rate of electron transfer is found to be independent of the speed of molecular reorientation. Electron transfer to the thionine singlet excited state from DNA obtained from calf thymus, salmon testes, and the bacterium, micrococcus luteus (lysodeikticus) containing different fractions of G-C pairs, has also been studied. Using a Monte Carlo model for electron transfer in DNA and allowing for reaction of the dye with the nearest 10 bases in the chain, the distance dependence scaling parameter,, is found to be 0.8 +/- 0.1 Angstrom(-1). The model also predicts the redox potential for guanine dimers, and we find this to be close to the value for isolated guanine bases. Additionally, we find that the pyrimidine bases are barriers to efficient electron transfer within the superexchange limit, and we also infer from this model that the electrons do not cross between strands on the picosecond time scale; that is, the electronic coupling occurs predominantly through the pi-stack and is not increased substantially by the presence of hydrogen bonding within the duplex, We conclude that long-range electron transfer in DNA is not exceptionally fast as would be expected if DNA behaved as a "molecular wire" but nor is it as slow as is seen in proteins, which do not benefit from pi-stacking.