The Chinese mainland government deliberately opened to visit Taiwan in recent years, the number of Chinese tourists increased year by year from 329,204 in 2008 to 4,184,102 passengers in 2015, but since 2016 due to political factors, the number of Chinese tourists began to decrease. In such a complex and volatile cross-strait situation, the traditional statistical analysis based on past experience and probability will inevitably make it difficult to predict the risky future. In order to make the Taiwan tourism industry in the shortest possible time, put forward effective countermeasures, in view of the future development. The prediction of the efficiency of Gray prediction model is very meaningful under limited and uncertain data. In this study, we use the gray theory to construct the forecasting of the dynamic situation under the uncertain environment and the incomplete information, using the two groups of mainland tourists from January to April in 2016 and from May to August in 2016 as samples. GM (1,1) and Gray Verhulst model is the more common choice. According to this study to predict the actual control of two models to the number of visitors to Taiwan tourists, the GM (1,1) model is more accurate than the Gray Verhulst model in terms of the number of tourists. Whether the errors or the number of future estimates of the month, GM (1,1) model can be based on the information to do the immediate response and anticipation.