To calibrate a CRT color monitor, several assumptions are often made about its performance, one of the most accepted being constant channel chromaticity. However, when the measurements are taken in a extended field of CRT screen, where it is assumed that only one primary phosphor is excited, the relative spectral radiance of the other primary phosphors that are contained in that field also contributes, if the background luminance is not zero. This contribution is greater at low levels of RGB channel stimulation and would invalidate the results obtained with calibration methods based on the hypothesis of constancy of chromaticity coordinates of RGB channels. We have studied this effect on the calibration of the CRT color monitor and, as a result, we propose a calibration that takes this effect into account. To evaluate the accuracy of the method, we have calculated color differences between the chromaticity that we wanted to reproduce and the one displayed with the CIELUV color-difference formula, and the color tolerances with the line element of MacAdam ellipses.