All coelenterate fluorescent proteins cloned to date display some form of quaternary structure, including the weak tendency of Aequorea green fluorescent protein (GFP) to dimerize, the obligate dimerization of Renilla GFP, and the obligate tetramerization of the red fluorescent protein from Discosoma (DsRed). Although the weak dimerization of Aequorea GFP has not impeded its acceptance as an indispensable tool of cell biology, the obligate tetramerization of DsRed has greatly hindered its use as a genetically encoded fusion tag. We present here the stepwise evolution of DsRed to a dimer and then either to a genetic fusion of two copies of the protein, i.e., a tandem dimer, or to a true monomer designated mRFP1 (monomeric red fluorescent protein). Each subunit interface was disrupted by insertion of arginines, which initially crippled the resulting protein, but red fluorescence could be rescued by random and directed mutagenesis totaling 17 substitutions in the dimer and 33 in mRFP1. Fusions of the gap junction protein connexin43 to mRFP1 formed fully functional junctions, whereas analogous fusions to the tetramer and dimer failed. Although mRFP1 has somewhat lower extinction coefficient, quantum yield, and photostability than DsRed, mRFP1 matures >10 times faster, so that it shows similar brightness in living cells. In addition, the excitation and emission peaks of mRFP1, 584 and 607 nm, are ~25 nm red-shifted from DsRed, which should confer greater tissue penetration and spectral separation from autofluorescence and other fluorescent proteins.