In a manufacturing or industrial process, reducing the variability of a systems and products is essential to increase yield and quality of the products. Statistical process control is a power collection of problem-solving tools useful to increase yield and quality of products through the reduction of variability. Traditionally average run length (ARL) is used to measure for the performance of statistical process control charts using integral equation, Markov chain approach and simulation studies. In this paper, an alternative to these methods a neural network approach for monitoring the process mean were proposed to examined the ARL performance of cumulative sum (CuSum) control chart. The results showed that the average run length (ARL) performance of CuSum control charts using neural network slightly outperforms than traditional ARL performance of CuSum control charts.