Understanding signal and noise quantities in any practical color imaging system is critical. Often noise quantities are assumed to be independent of the signal, independent of color channel and either uniform or Gaussian additive. These simplistic models are not realistic and there is a need for accurate noise models in order to design optimal color imaging systems. Different noise characteristics between the individual color channels should be taken into account when developing demosaicing methods, noise reduction algorithms and other image processing tasks. The choice of color space in which to operate on a color image is also dependent on the noise characteristics of the sensor. It is also possible to optimize the filters in the CFA (color filter array) based on knowledge of the sensor noise. We describe a noise model for a modern APS CMOS detector and a number of noise sources. A method for characterizing the noise sources given a set of dark images and a set of flat field images is outlined. The noise characterization data is used to simulate dark images and flat field images. The simulated data is a very good match to the real data thus validating the model and characterization procedure.