The present methods for risk analysis, such as sensibility analysis and “decision tree” analysis, have the following disadvantages such as assumption of factor independence, subjective factors in value taking and incomplete range of factor value. A new approach to risk analysis was studied with the introduction of conditional probability to correlate the risk factors. A probability distribution function derived by calculating a series of net present values of the project was used to decide the probability of risk. As a result, the theory and case study show that the improved method for risk analysis can effectively characterize the interactive relationship between risk factors, and can also achieve a better estimation of project risk. The method can offer preference for the risk analysis of non-agricultural projects and provide decision-making basis for investors and administrators.