The prediction of color appearance using the RLAB color space has been tested for a variety of viewing conditions and stimulus types. These tests have shown that RLAB performs well for complex stimuli and not-so-well for simple stimuli. This article reviews the various psychophysical results, interprets their differences, and describes evolutionary enhancements to the RLAB model that simplify it and improve its performance.