We conducted an experiment with actual windows in the rooms to examine the applicability of the formulae validated with simulated window to actual windows. Generally subjective evaluation in this experiment had the same tendency as our previous results with simulated window. However, both DGI and UGR were considered insufficient to predict glare sensation in all conditions. Inconsistency was shown in the case of very large source and in the difference in window orientation. Overating background luminance in these formulae must contribute to this reason. From the multiple regression analysis it was found that "the effect of total amount of light" must not be neglected in the case of a large light source.