Estimation of the spectral characteristics of color image sensors is an important problem. The effective spectral response of a color imaging system may vary from the color filter specification for a variety of reasons. However, color correction and color constancy algorithms require precise characterization of the color filter response in order to attain accurate color reproduction. Also, camera systems need to be qualified and calibrated quickly in a manufacturing environment. CMOS image sensors, which are becoming more popular, complicate the problem, since noise levels on CMOS sensors have yet to equal their CCD counterparts. For these reasons, a fast and reliable characterization of the color sensor response is needed. This paper presents an approach to the problem using numerical regularization techniques. In particular, Tikhonov regularization is used to determine the spectral response of a color image sensor given the digital output codes from an image with known reflectances under a known illuminant.