This study develops a scenario-based multi-objective optimization framework for the geometric design of a real-world labyrinth spillway by coupling empirical hydraulic relationships with two evolutionary algorithms: Multi-Objective Particle Swarm Optimization (MOPSO) and Multi-Objective Imperialist Competitive Algorithm (MOICA). The optimization problem was formulated to maximize discharge capacity and minimize concrete volume while satisfying hydraulic, geometric, structural, and design-discharge constraints. Three design scenarios with different levels of geometric freedom were considered to evaluate the influence of constraint severity on the hydraulic response and attainable optimized solutions. The framework was applied to the Ute Dam labyrinth spillway as a benchmark case. Results from ten independent runs showed that MOPSO outperformed MOICA in convergence stability, objective-space coverage, and robustness. MOPSO achieved a higher mean Hypervolume value of 0.721 compared with 0.662 for MOICA, representing an improvement of approximately 8.9%. It also reached Hypervolume stabilization earlier, after 42 iterations compared with 57 iterations for MOICA, and showed a lower Hypervolume standard deviation of 0.009 compared with 0.017. The scenario-based results demonstrated that the most flexible design case produced the most favorable volume–discharge response, whereas excessive geometric enlargement did not necessarily improve hydraulic performance. Increasing concrete volume from 6000 to 9500 m³ reduced discharge capacity by approximately 3.52–4.41%, and the volume-normalized hydraulic productivity decreased by about 39%. These results indicate that balanced geometric proportions are more important than structural enlargement alone.