<p>In current engineering practice, the vibration serviceability performance is usually checked after the safety design is finished. Such process does not guarantee the optimized design, especially for those large-span truss structures, where the vibration serviceability performance is the controlling factor that dominates the design process. The randomness of the human-induced loads and structures significantly affects the design results. To address such problem, a dynamic-reliability-based topology optimization (DRBTO) method for the vibration serviceability design of large-span truss structures under random human-induced loads is proposed, which employs the probability density evolution method (PDEM) for dynamic reliability analysis and the density approach for solving topology optimization. To reduce the calculation of multiple order force vectors, the conjugation of the force vectors is utilized to reduce half of the calculation in finite element analysis and sensitivity analysis. Unlike other types of dynamic loads, the load positions of human-induced loads also have strong randomness, so the load positions are simulated as uniformly distributed random variables. To better implement the optimization, sensitivity analysis function is adopted for efficient calculation of the structural response sensitivity. Gray-scale operator is adopted to avoid the occurrence of intermediate density. Numerical results with different volume fractions and acceleration thresholds show that the proposed method could give better design results from the perspectives of structural responses and dynamic probability, thus demonstrating the effectiveness and stability of this optimization method.</p>