This paper develops an approach to minimize the number of process tooling adjustments and deliver an acceptable fraction of non-conforming products based on given product quality specification limits in assembly processes. A linear model is developed to describe the relationships between product quality and process tooling locating positions. Based on the model, the process mean shifts of tooling locating positions are estimated for both deterministic and stochastic cases by using the least-square estimation or linear mixed model estimation, respectively. A simultaneous confidence interval is obtained to construct the estimation region of a process mean shift under the given false alarm rate. Furthermore, a tooling adjustment strategy is proposed to determine when the process adjustment is essentially needed in order to ensure an acceptable fraction of non-conforming units based on the given product quality specification limits. Finally, a case study is conducted to illustrate the developed methodology by using a real-world autobody assembly process. Copyright © 2010 John Wiley & Sons, Ltd.