Modeling of fluorescent color mixing is studied by applying regression analysis to the measured bispectral data. A sample set was prepared by using three fluorescent paints and different mixing combinations. The accuracy of estimations was evaluated by calculating the color difference ΔE in the CIELAB color space and Euclidean distance of spectra. Even though the estimation errors were considerably high, the best results overall was obtained using a sparse training set and of third order terms in regression.