Many factors are involved in the prediction of College Entrance Examination (CEE) aspiration which is a non-linear classification problem. We proposed a CEE aspiration prediction approach based on support vector machine learning algorithm. Firstly, CEE score and ranking in all subjects, the number of college admission plan and relevant data of the latest two years are collected and a training set is formed. Secondly we analyze the influential factors of CEE admission, and there are 14 features, such as score, score sorting, the lowest admission fractional lines of all batches, the number of enrollment plans of all batches in all levels of colleges and universities and school enrollment plans .And feature extraction is implemented on the two years' data to obtain the training staff for prediction, then the machine learning algorithm of Support Vector Machine is used to train the decision-making process of CEE aspiration and the analytical model for prediction is established. Finally, the admission data of 2009 and 2010 partial examinees is applied on prediction experiment. The result shows that the proposed method performs a very good effect, the prediction accuracy reaches 90%, giving very favorable guidance to examinees for aspiration filling.