At this research, automobile MPG (Miles Per Gallon) forecasting model developed through multiple linear regression and artificial neural network. Mahalanobis Distance(MD) outlier detection method used for the purpose increase forecasting ability of mathematical model. The detailed procedures described for calculating MD. The effects of outlier are compared with different data size for neural network training. MD outlier detection method successfully removed existence of outlier and improved the neural network training results and MPG estimating performance.