With the rapid development of the economic, the income of Chinese residents is increased significantly, however the level of urban-rural income gap has also been increasing. Income inequality is growing and it has already aroused extensive attention of the whole society. Based on the data of income gap per capita between rural and urban areas from 1978-2008 in China, it fits the regressive curve by trend graph and establishes an time series model of ARMA by the residual sequence, furthermore analysis, identification, estimation and test of the model parameters were conducted to establish a suitable forecast model of China's urban-rural income gap. The outcome proves that the model takes on the traits of simple shortcut, high forecasting precision and adaptation to actual needs. The prediction indicates that China's urban-rural income gap is widening, the state should take positive measures to control the situation. State should increase investment in agriculture, protect the interests of farmers and raise the income of peasants to maintain social stability.