The enterprises will face up with many uncertainties when they make the project investment risk decisions. The uncertainties can lead to the deviation between the actual result and the expectant result, reduce the return rate of investment and extend the investment payback period. The key of making scientific decisions is to measure and analyze the project investment risk accurately. Traditional methods about the project investment risk decision include three layers: Payback Period Method without considering the monetary value of time; Net Present Value Method (NPV) and Internal Rate of Return Method (IRR) with considering the monetary value of time; Risk-adjusted Discount Rate Method and Certain-equivalent Method that combine the monetary value of time with the risk. The first layer method cannot be treat as the main method, because it does not consider the monetary value of time, the risk and the revenue after investment recovery period. Because of the same reason, the second layer methods are not consistent with the actual situation. The third layer methods make more progress than the two previous layers methods, but they also have some defects. Basing on the different degree project risks, the Risk-adjusted Discount Rate Method uses the suitable discount rate to calculate the NPV. This method will increase the risk lately, because it mixes the monetary value of time with the risk value. The Certain-equivalent Method uses different factors to adjust uncertain cash flow to certain cash flow and then calculate NPV. But the most difficult thing is to confirm the suitable factor. Normally the enterprises use one method to make decisions, it is very incomplete. Basing on the prior analysis, this paper will construct an index system, including non-risk-adjusted NPV, PI, PP. comprehensive standard deviation and Coefficient of variation. Then using the Entropy Weight Evaluation Method to confirm the best program. This will enhance the science and feasibility of the investment risk decisions.