<p>The present work proposes a hierarchical multi-objective optimization (HiMOO) framework for the hybrid variables design of thin-walled tubular deployable composite booms (TDCBs). The framework addresses a complex mixed-variable problem involving discrete stacking sequences and continuous geometric parameters (ply thicknesses, radius, central angle), aiming to minimize structural weight and maximize winding torque under constraints of failure index, fundamental frequency, and manufacturability. The HiMOO framework operates through two hierarchical stages: the internal stage optimizes continuous variables for fixed stacking sequences via Non-dominated Sorting Genetic Algorithm II (NSGA-II), while the external stage employs a novel variable-aggregated dominance order ranking (VADOR) method to refine discrete sequences. Analytical models for winding torque and failure index are developed to efficiently assess fitness values. To alleviate computational costs, finite element simulations combined with a multi-point approximation method construct surrogate models for fundamental frequency analysis. The process initiates from a ground-based laminate structure, enabling flexible adaptation to arbitrary stacking sequences. Two case studies validate HiMOO’s superiority over single-stage approaches. The first benchmark demonstrates expanded Pareto fronts with improved trade-offs between torque and bending stiffness when relaxing ply constraints. The second engineering case reveals that HiMOO-driven designs significantly reduce weight with the same winding torque level compared to conventional methods, with near-optimal sequences favoring 0°/90° plies to balance stiffness and failure resistance. The hierarchical strategy effectively decouples discrete–continuous variable interactions, offering computational efficiency and broader design exploration. This work provides a systematic methodology for deployable composite structures and general mixed-variable optimization problems.</p>