changes in a sensor domain are translated into a change in energy-transfer efficiency between donor and acceptor fluorescent domains, which is detected by a change in the ratio of donor and acceptor emission. This ratiometric response is independent of the sensor concentration, which is an important advantage of FRET-based sensors. In practice, however, most FRET-based sensors display only a relatively small difference in emission ratio upon activation. Improvement of these ratiometric changes has been recognized as an important prerequisite for use of these sensor systems in high-throughput applications based on fluorescence plate readers and fluorescence assisted cell sorting (FACS). [14, 15] Recently a pair of CFP (cyan fluorescent protein) and YFP (yellow fluorescent protein) variants, CyPet and YPet, respectively, have been reported that were optimized for FRET through a process of directed evolution. [16] When incorporated in a protease sensor, a 20-fold change in emission ratio was observed upon cleavage of a flexible peptide that linked CyPet and YPet, compared to only a fourfold change for the same construct with enhanced CFP (ECFP) and enhanced YFP (EYFP) domains. However, the mechanism behind their remarkable FRET properties has remained unclear. A total of eighteen mutations were introduced in the course of their development, many of which were at the exterior of the protein, at a large distance from the fluorophore. Moreover, no large differences in quantum yield or extinction coefficient were reported; this suggests that the photophysical properties of the fluorescent proteins were not significantly altered. We therefore hypothesized that the increase in FRET observed for CyPet and YPet could be due to an enhanced tendency to interact when connected by a peptide linker. The parent green fluorescent protein (GFP) has a known tendency to dimerize, [17] and analysis of the mutations in YPet have identified two residues, S208F and V224L, that are present at the dimer interface, as shown by the X-ray structure of the GFP dimer. Here, we show that A of just these two mutations in both fluorescent domains of ECFP–linker–EYFP constructs results in a fourfold increase in the EYFP-to-ECFP emission ratio, which yields a 16fold change in emission ratio upon protease cleavage of the peptide linker (Figure 1). Additional biophysical evidence is provided, which shows that the mutations indeed result in A of an intramolecular complex.