<p>Needle-punching is an effective route for introducing Z-direction fibers into preforms for carbon/carbon (C/C) composites, but excessive needling can also damage the continuity of in-plane fiber bundles. To address the resulting multi-objective conflicts among mechanical properties, this study developed a collaborative optimization strategy integrating response surface methodology (RSM), multi-objective hippopotamus optimization (MOHO), and multi-attribute decision-making. Based on the Box–Behnken Design, second-order regression models correlating needling depth, needling density, and carbon cloth areal density with tensile, compressive, and flexural strengths were constructed. The results showed that these properties exhibited different sensitivities to the process parameters, reflecting the trade-off between Z-direction reinforcement and in-plane fiber damage and identifying a compromise process window rather than a single-property optimum. For the present optimization problem, MOHO was used to generate the Pareto non-dominated solution set and, compared with the representative baseline NSGA-II, yielded a more uniformly distributed Pareto solution set and a better compromise-ranking result. Experimental verification demonstrated that under the optimized combination (needling depth 17 mm, needling density 22 punches/cm², and carbon cloth areal density 377 g/m²), the comprehensive mechanical properties achieved an optimal balance with a maximum prediction error of 5.85%. The proposed strategy provides a practical route for balancing competing mechanical-property requirements in needle-punched C/C composite preforms.</p>