In many engineering applications, both random and interval variables exist. The mixture of both types of variables has been dealt with in robust design for single-disciplinary systems. This work focuses on robustness assessment for multidisciplinary systems with both random and interval variables. To alleviate the intensive computational demand, a semi-second-order Taylor expansion method is proposed to evaluate robustness. A performance function is approximated with linear terms of random and interval variables, as well as the interaction terms between the two types of variables. Then the maximum and minimum standard deviations of the performance function are computed. With the proposed method, the impact of both random and interval variables on the system robustness can be evaluated efficiently.