Fraudulent financial statement will mislead the due diligence，thus making investors suffer losses. It is important to know how to detect the fraud in the early stage of a project. So institutions that need to do surveys could keep track of the characteristics of the fraudulent financial statement，and stop the project in time if fraud occurs. Based on the enterprise sample from Jan.2000 to Jan.2018 that received regulatory penalty for profit fraud，this paper builds a financial statement fraud detecting model using factor analysis and logistic model. We find asset quality，earnings and the stability of cash flow is important to detect the fraud. Also，the change of audit firm，the net profit after deducting non regular profit and loss is meaningful for detecting.