Assessment on branch importance under cut-off power flow helps to analyze the size of system disturbances caused by breaking of different branches. This article has proposed four factors which affect the branch importance based on cut-off power flow. After a branch importance assessment form is formed, according to the actual characteristics of the data sheet, this article has put forward an attribute weight assessment algorithm based on EM clustering, and has verified it with an example of a 38-node power system. It can be found from the weight vector of an attribute that branch importance is mainly determined by DifV, DifGenPower and DifBrPower. Through the comparison of the original score sheet and the assessment form, we can find that the ultimate branch importance sequencing is consistent with the above conclusion, which thus has proved the correctness of the algorithm.