This work proposes a conservative manifold least-squares Petrov–Galerkin (LSPG) projection approach for model reduction of steady hypersonic aerodynamics. Hypersonic flows exhibit complex physics including shock waves, boundary layers, and high-temperature thermochemical effects, making them computationally expensive to simulate. The proposed method combines the conservative formulation for finite-volume discretizations with manifold LSPG projection to enable accurate and efficient reduced-order models for hypersonic aerodynamics. Numerical experiments demonstrate significant computational speedups while maintaining accuracy for engineering-relevant quantities of interest.