Cumulative sum (CUSUM) control charts are very effective in detecting special causes. In designing a CUSUM chart, it is important to know the average run lengths (ARL's) corresponding to various possible choices of the mask parameters. This paper provides a relatively simple yet very accurate (typically within 3%) approximation for the ARL of a CUSUM chart, both when the process is in control and when it is out of control. Examples are given showing how the approximation can be used to evaluate the ARL for specific parameter values, to find values that give a desired ARL, and to evaluate the out-of-control ARL's of location and scale CUSUM charts.