As the intensity of urban underground space development increases, closely spaced large-span overlapping tunnel groups have been widely implemented in mountainous urban transport infrastructure. However, the superposition of complex geological conditions and multi-tunnel interactions has made stress and deformation control during construction increasingly challenging. To address the difficulty of simultaneously achieving construction safety and efficiency, group excavation techniques and bench-length optimization for closely spaced large-span overlapping tunnel groups were investigated. Based on an overall “down-first, then-up” strategy, a “station-first three-bench–coordinated upper–lower bench excavation in running tunnels (Scheme 2)” method was proposed. Using staged surface settlement as the controlling index, field monitoring data were compared with numerical simulation results, and consistent evolution trends were observed, with errors ranging from 10.1% to 16.8%. Regarding displacement control around the tunnel perimeter, a slightly larger crown displacement was produced under Scheme 2 (maximum difference of 2.59 mm); nevertheless, the smallest clearance convergence was obtained, and a slower and more uniform overall settlement rate was achieved, indicating improved construction-control performance. In terms of stress-field evolution, larger minimum principal stresses were generally generated under Scheme 3, suggesting that the complicated support system constrained the self-bearing capacity of the surrounding rock. By contrast, under Scheme 2, the inherent self-supporting capacity of the surrounding rock was allowed to be more fully mobilized while safety requirements were satisfied, which is beneficial to the long-term stability of the tunnel group. Overall, this method reduces construction-process interference and balances construction safety with efficiency. Sensitivity analysis further indicated that the bench length of the main tunnel exerted a significant influence (Factor C: F-ratio = 4.667), whereas the bench length of the three longitudinal tunnels showed a secondary effect (Factor B: F-ratio = 1.734) and the ramp tunnel bench length exhibited only a minor influence (Factor A: F-ratio = 0.329). Therefore, priority should be given to adjusting the bench length of the main tunnel (Factor C) during optimization to improve the target performance indices.