Assessment and optimization of illumination are difficult issues in automatic visual inspection. After analyzing the images of visual inspection of printed circuit boards and their pixel value distribution in three-dimensional color space, an illumination assessment method is proposed. This merit function is based on quantitative analysis of Euclidean distance of color in red-green-blue (RGB) space, fluctuation extent of pixel value and deviation of defect color. The experimental results show that this method is suitable for evaluating the illumination effect and complies with subjective feeling of human visual system. It is also shown that the precision is better than 1%. An illumination optimization experiment is performed and the optimum lighting effect is obtained based on the merit function. This method can be applied to other visual inspection areas, such as color print quality inspection and plastic product defects inspection.