Abstract: Aiming at the difficulty of balancing the minimization of main arch bending moments with complex structural requirements in arch bridge design using the traditional pressure line theory, a multi-objective shape optimization method for arch axes based on the non-dominated sorting genetic algorithm (NSGA-II) was proposed in this paper. Based on a 30 m-span hingeless arch bridge in Chizhou City, Anhui Province, a parametric geometric model was established using the catenary equation. The bending moment distribution function and the bending strain energy calculation model under distributed loads were derived. Subsequently, a dual-objective optimization model was constructed, aimed at minimizing the bending strain energy of the main arch while precisely controlling the slope at the arch foot. The Pareto optimal solution set was searched using the non-dominated sorting and elite retention strategies of the NSGA-II algorithm. The results show that the optimized arch axis coefficient is adjusted from an initial value of 2.000 to 2.484, leading to a 13.40% reduction in the bending strain energy of the main arch ring while ensuring the arch foot slope remains within a reasonable structural range. Compared with the classic five-point coincidence method, the NSGA-II optimization scheme exhibits a slight compromise in local bending indicators, but the total strain energy is reduced by approximately 7.27%, achieving a scientific trade-off between axial compression and bending performance. Parameter sensitivity tests and stability verification demonstrate that the algorithm converges stably within 100 generations, with a coefficient of variation of only 1.56% across independent runs, indicating good robustness. This study reveals the intrinsic correlation between the arch axis coefficient and the structural mechanical response, providing a quantitative pathway and theoretical support for the type selection design of small- and medium-span arch bridges under multi-objective trade-offs.