Through analyzing the evaluation criteria of virtual enterprise knowledge sharing risk early-warning system, a hierarchical structure evaluation attribute model is proposed in this paper. Then, utilizing the basic theory and method of rough sets and grey theory, a virtual enterprise knowledge sharing risk early-warning model framework structure based on rough sets is established. In order to optimize attributes, the heuristic attribute reduction algorithm based on discernable matrix is put forward. On this basis, virtual enterprise knowledge sharing risk is assessed with the use of grey theory. Finally an example is used to verify the effectiveness and the practical value of the method, the evaluation result of risk early-warning coincides with the actual result.