The Jameson and Hurvich opponent-colors model of hue and saturation was tested for spectral and non-spectral lights. Four observers described the color of lights by scaling hue and saturation. The lights ranged from 440 to 640 nm and consisted of five purities: 1.0, 0.80, 0.60, 0.40 and 0.20. Admixtures of monochromatic and a xenon-white light yielded the different colorimetric purities. For each subject, chromatic response functions were measured by the method of hue cancellation at each purity, and an achromatic response function was measured by the method of heterochromatic flicker photometry for spectral lights. Chromatic response functions measured for a particular purity and the achromatic response function were used to predict hue and saturation for that purity. The model successfully predicted hue at each level of purity, but failed to predict precisely the Abney effect. The model made relatively poor predictions of saturation, tending to overestimate short-wave lights and underestimate long-wave lights. An additional experiment found that stimulus parameters that favor rod contribution weaken the model's predictions of saturation, while stimulus parameters that do not favor rod contribution improve the model's predictions of saturation.