Abstract: The optimization design of Exponential Weighted Moving Average(EWMA) control chart for nonnormal data is addressed. A method that combines Taguchis loss function and the probability distribution of the random process mean shift is proposed. By using the Burr distribution to approximate various nonnormal distributions, the parameter of the EWMA control chart for nonnormal data is determined through minimizing the overall mean value (ML) of Taguchis loss function. Comparative studies show that the proposed MLEWMA control chart for nonnormal data is significantly superior to the statistical EWMA control chart for nonnormal data in terms of the overall loss of ML.