Existing color quality indices for light sources provide broad information about different dimensions related to color quality. Color fidelity, harmony, and gamut area are concepts related to these indices, and industry requests this information. For the last few years, LED light sources have been widely used at home and at work, and now a color rendering index that solves the problem of underestimation of this type of light source is needed to provide a score of subjective assessments made by real observers related to color fidelity. In this work, this problem has been studied in two ways, theoretically and experimentally, and the results show discrepancies between the hue composition calculated theoretically using a color appearance model and hue composition evaluated by real observers. These discrepancies could originate from divergences in the color fidelity score and the subjective evaluation of the naturalness of a scene.