Photolithography in the semiconductor fabrication process is the core stage that determines the quality of semiconductor chips. The fabrication process is a batch process that causes variation in the quality of chips; thus, uniformity has always been an important goal of the process. This research is a case study on optimizing the photolithography stage to improve uniformity and the target achievement of critical dimension, a quality measure of semiconductor chips. The case study finds the optimal setting of input variables in photolithography by applying multivariate normal linear modeling with operational data obtained by sampling during manufacturing. The predicted performance of the optimal setting is found to be close to the limit of improvement estimated based on the model. For practitioners, several issues that can be considered for better optimization are also provided. Copyright © 2010 John Wiley & Sons, Ltd.