Engineered luciferase-luciferin pairs have expanded the number of cellular targets that can be visualized in tandem. While light production relies on selective processing of synthetic luciferins by mutant luciferases, little is known about the origin of selectivity. The development of new and improved pairs requires a better understanding of the structure-function relationship of bioluminescent probes. In this work, we report a biochemical approach to assessing and optimizing two popular bioluminescent pairs: Cashew/d-luc and Pecan/4-BrLuc. Single mutants derived from Cashew and Pecan revealed key residues for selectivity and thermal stability. Stability was further improved through a rational addition of beneficial residues. In addition to providing increased stability, the known stabilizing mutations surprisingly also improved selectivity. The resultant improved pair of luciferases are >100-fold selective for their respective substrates and highly thermally stable. Collectively, this work highlights the importance of mechanistic insight for improving bioluminescent pairs and provides significantly improved Cashew and Pecan enzymes which should be immediately suitable for multicomponent imaging applications.