Forty-six young adult observers with normal color vision (plus five from an earlier study) made a series of color matches using a new LED-based, multi-wavelength visual trichromator. Thirteen LED lights of different wavelengths were combined to produce 11 triplets of lights that observers were asked to match to a white reference light of 7500 K over visual angles of either 2° or 10°. Matches were initially made by asking observers to adjust the intensities of the three lights making up each triplet. As the experiment progressed, a more intuitive matching procedure was developed. By transforming the triplet of lights into CIELAB space, observers adjusted colors using lightness (L*), redness-greenness (a*), and blueness-yellowness (b*) to make the match with white. The new procedure proved easier for observers and reduced the inter- and intra-observer variability. Given that each of the 11 matches to the reference white for a given observer (obtained by either method) should produce identical L-, M- and S-cone excitations, we were able to use the matches to infer the individual cone spectral sensitivities for each observer and thus estimate the range of individual differences across our 51 observers. By applying a model of the CIEPO06 standard LMS observer, the photopigment, macular and lens optical densities and the L- and M-cone photopigment spectral shifts that best equated the three-cone excitations across the 11 matches were found for each observer. The individual differences were consistent with the CIEPO06 observer except for a 3 nm shift of the M-cone photopigment to longer wavelengths and a slightly denser 2-deg macular pigment density.