Abstract: For the multi-component isotope separation, there are many different cascades capable of fulfilling a specified separation task. The total flow of cascade is a significant performance target for the optimal design of cascades. The optimal designs of three types of cascades, namely, matched-R cascade, matched-X cascade and quasi-ideal cascade, were investigated by numerical approach with the help of virtual components, and their separation performances were compared for accomplishing the same separation task. The results show that in most cases, quasi-ideal cascade and matched-R cascade perform better, and the former is slightly superior to the latter. The matched-X cascade is the worst of the three types. Therefore, the optimal design of cascades for multi-component isotope separation should take quasi-ideal cascade as the first choice.