This paper presents a new, computationally efficient robustness measure for robust geotechnical design (RGD) faced with a lack of statistics of statistical parameters of variables, based on reliability analysis of the system response. Considering the computational coupling between the new robustness measure and the reliability of the system, the existing framework of the reliability-based RGD approach is simplified. To illustrate the simplified and improved RGD approach, the design of rock wedge slopes is considered herein as an example. For the robust design of rock wedge slopes with combined failure modes, an N-dimensional equivalent method is adopted to ensure the feasibility and accuracy. Taking design robustness and cost efficiency as two objectives, robust design optimization is conducted while the design constraint (safety requirement) is enforced. The robust design optimization yields a set of design solutions rather than a single best solution because of the conflicting relationship between the two objectives. Based on the developed Pareto front, the most preferred designs can be specified. This design approach is illustrated through the design example of a rock wedge slope, and compared with the most preferred designs of reliability-based design (RBD), some suggestions of this robust design approach in geological engineering are presented.