Abstract: The coupling between the internal and external flows of high-speed aircraft is tight and complex, and airframe/propulsion integrated design is an important approach to improving overall performance. The construction of adjoint equations based on an upwind scheme and turbulence variation is studied. The left-hand-side terms of the adjoint equation, including the inviscid term, viscous term, and boundary condition Jacobian, and the right-hand-side terms, including the variational forms of total pressure recovery coefficient and thrust coefficient, are systematically derived. Based on the discrete adjoint method, free-form deformation (FFD) parameterization, and sequential quadratic programming (SQP) algorithm, the adjoint optimization framework is constructed, and integrated airframe/propulsion optimization is carried out for a typical air-breathing high-speed vehicle, with the lift-to-drag ratio, inlet total pressure recovery coefficient, and nozzle thrust coefficient as the objectives. The free-stream Mach number is 6, the angle of attack is 4°, and the Reynolds number is 2.26×10⁶. Multiblock structured grids are used, and 241 design variables are employed. The optimization is performed on a 64-core server and converges after 15 iterations within 16 h. The results show that the lift-to-drag ratio is increased by 2.1%, the total pressure recovery coefficient of the inlet is increased by 4.8%, and the thrust coefficient of the nozzle is increased by about 0.5%, which verifies the robustness, effectiveness, and efficiency of the method established in this paper.