In a color matching experiment, the colors of two stimuli might match perfectly to one observer. However, to another, those same two stimuli might not match depending on the spectral characteristics of the two stimuli and each of the observers' visual responses. This phenomenon, that any two observers perceive color matches differently, is termed observer metamerism. Only recently have differences in the factors that affect the spectral response of the human eye been quantified to a degree such that observer metamerism can be realistically modeled over a representative population of human observers. From the statistics of these factors, a large, representative group of such observerseach said to be metameric to a standardwas created in terms of LMS cone spectral responses. Using analysis of variance and error propagation, as suggested by Nimeroff, this sample set was reduced to a complete colorimetric description that includes mean color matching functions along with variance and covariance functions. This simplified representation is then applied to creating arbitrary sets of color matching functions, such as the CIE 1931 and 1964 observers, and their covariance functions to allow the practical characterization of observer-metameric mismatch gamuts for any nominally metameric stimulus pair. (c) 2015 Wiley Periodicals, Inc. Col Res Appl, 41, 115-124, 2016