Since air drag is proportional to the square of the speed, it is expected that reducing air drag will significantly improve fuel efficiency for on-highway trucks and buses, which are often driven at high speeds. Therefore, the purpose of this study is to propose an optimization method for vehicle shape to drastically reduce aerodynamic drag in heavy-duty vehicles. Using NSGA-II, one of a genetic algorithm, the overall vehicle shape was optimized with drag coefficient (CD) and lift coefficient (CL) values as objective functions and design variables as parameters in a total of 13 locations. Among the Pareto solutions, an 86% reduction in CD was achieved compared to the base shape when the CD value was the lowest. Since the CL value remains low with this shape, it can be seen that driving stability does not deteriorate. Among the design variables in optimization, it was confirmed that the corner radius of the vehicle side was particularly effective in reducing the CD value. In addition, when optimizing only the cab shape, the optimal value for the front virtual angle was 30 deg., but in relation to the corner radius, the value for this optimized shape was around 40 deg. The CD value of a 1/20 model of the optimized shape was measured in a wind tunnel test and compared with the optimization results from the aforementioned numerical analysis. As a result, the CD value reduction effect of the optimization shape was confirmed in the wind tunnel test as well, demonstrating the validity of the optimization results described above. In addition, an investigation into the yaw angle dependency of the optimized shape revealed that adding a yaw angle provided a sailing effect that reduces CD. A program for calculating fuel consumption rates for heavy-duty vehicles was used to compare fuel efficiency when using the base shape and the optimization shape. The weighted average fuel economy in urban driving (JE05) and interurban driving (highway) modes was improved by approximately 21% using the optimization shape.