This study implemented multi-objective design exploration for an automatic transmission casing using genetic algorithm (GA) and data mining techniques. In general, real-world design problems are requested to be solved by considering all relevant disciplines simultaneously, which is called multidisciplinary design optimization (MDO). We often face the difficulty in balancing these different disciplines to satisfy required performance, such as durability, stiffness, noise, vibration, harshness, mountability, weight, and manufacturability considered in this study. In addition, we often have to improve development efficiency for shortening of the development period. Therefore, MDO has been widely implemented for taking improved design candidates according to required specifications and making a decision to fix conceptual design in an early stage. MDO is usually considered as the multi-objective optimization which aims to obtain the set of Pareto-optimal solutions. These solutions happen due to the trade-off relationships between competing objectives, and are useful to characterize the potential of a design product to be optimized. From the aforementioned points of view on MDO, this study applied GA for the MDO of the automatic transmission casing because of the superior solution searching capability in multi-objective optimization. Our design objectives were to minimize the casing weight and noise by changing the wall thickness. We divided the whole casing surface into 669 regions, and define the thickness in each region individually as the design variable in the present MDO. Since GA is computationally expensive due to the population-based search, this study combined GA with Kriging surrogate models to reduce time required for evaluating the searched solutions. Furthermore, during the optimization process, we analyzed the obtained solutions to see the trade-off relationships using data mining techniques. Finally, we obtained the shape of automatic transmission casing that was lighter than initial shape and satisfied required performance.