This study discusses the accuracy of daylight illuminance predicted by computation algorithms that were developed for the separate use of the sun and sky as a light source. Computer simulations based on the algorithms were conducted to predict daylight illuminance in office space. In order to examine the prediction accuracy of the algorithms, the prediction results were compared to the prediction results from the Radiance. The results of study imply that the prediction results by the algorithms were similar to those by the Radiance. Among all algorithms, the sun-matching method generated the least deviations from the prediction results by the Radiance. Compared to the one sky patch method, the four sky patch method generated closer results to those of the Radiance. The sky-matching method provided less deviation from the results of Radiance than the daylight coefficient approach. Linear regression models between the prediction results by the computation algorithms and the Radiance indicate that strong linear correlations existed between them. The prediction results based on the algorithms suggested in this study were statistically effective to predict the daylight illuminance in office.