As the efficiency measuring and analyzing is a crucial problem in modern enterprise management, this paper focuses on the problem of small and medium enterprise financing efficiency analyzing, and we proposed an improved DEA model based on stepwise regression to tackle this problem. Data envelopment analysis technology can calculate the efficiency of each decision making unit relative. Hence, DEA can provide an efficient frontier or envelop for all considered decision making units rather than fitting a regression plane through the center of the data. To solve the limitations in traditional DEA model, we propose an improved DEA model using the stepwise regression technology which has the ability to automatically select variables for the DEA model. To test the effectiveness of our proposed DEA model, 120 small and medium enterprises of China in 2010-2013 is chosen to construct a dataset, and this dataset are collected from the annual reports of sample firms. In particular, in this experiment, the input indexes we used are (1) Total assets of an enterprise, (2) Main business cost, (3) Asset liability ratio. On the other hand, the output indexes are (1) Rate of return on net assets, (2) Main business income, and (3) Total assets turnover. Combining all the experimental results, some suggestions to enhance the small and medium-sized enterprise financing efficiency in China are offered.