The neutral form of the chromophore in wild-type green fluorescent protein (wtGFP) undergoes excited-state proton transfer (ESPT) upon excitation, resulting in characteristic green (508 nm) fluorescence. This ESPT reaction involves a proton relay from the phenol hydroxyl of the chromophore to the ionized side chain of E222, and results in formation of the anionic chromophore in a protein environment optimized for the neutral species (the I* state). Reorientation or replacement of E222, as occurs in the S65T and E222Q GFP mutants, disables the ESPT reaction and results in loss of green emission following excitation of the neutral chromophore. Previously, it has been shown that the introduction of a second mutation (H148D) into S65T GFP allows the recovery of green emission, implying that ESPT is again possible. A similar recovery of green fluorescence is also observed for the E222Q/H148D mutant, suggesting that D148 is the proton acceptor for the ESPT reaction in both double mutants. The mechanism of fluorescence emission following excitation of the neutral chromophore in S65T/H148D and E222Q/H148D has been explored through the use of steady state and ultrafast time-resolved fluorescence and vibrational spectroscopy. The data are contrasted with those of the single mutant S65T GFP. Time-resolved fluorescence studies indicate very rapid (<1 ps) formation of I* in the double mutants, followed by vibrational cooling on the picosecond time scale. The time-resolved IR difference spectra are markedly different to those of wtGFP or its anionic mutants. In particular, no spectral signatures are apparent in the picosecond IR difference spectra that would correspond to alteration in the ionization state of D148, leading to the proposal that a low-barrier hydrogen bond (LBHB) is present between the phenol hydroxyl of the chromophore and the side chain of D148, with different potential energy surfaces for the ground and excited states. This model is consistent with recent high-resolution structural data in which the distance between the donor and acceptor oxygen atoms is : <= 2.4 angstrom. Importantly, these studies indicate that the hydrogen-bond network in wtGFP can be replaced by a single residue, an observation which, when fully explored, will add to our understanding of the various requirements for proton-transfer reactions within proteins.