Nowadays, the rising interest in using renewable energy for thermal power generation has led to radical changes in steam turbine design practice and operability. Modern steam turbines are required to operate with greater flexibility due to rapid load changes, fast start-up, and frequent shutdowns. This has given rise to great challenges to the exhaust hood system design, which has a great influence on the overall turbine performance converting the kinetic energy leaving the last stage of low-pressure turbine into static pressure. The radial hoods are characterized by a complex aerodynamic behavior since the flow turns by 90 deg in a very short distance and this generates a highly rotational flow structure within the diffuser and exhaust hood outer casing, moreover, the adverse pressure gradient can promote the flow separation drastically reducing the hood recovery performance. For these reasons, it is fundamental to design the exhaust system in order to ensure a good pressure recovery under all the machine operating conditions. This paper presents a design of experiment (DOE) analysis on a low-pressure steam turbine exhaust hood through computational fluid dynamics (CFD) simulations. A parametric model of an axial-radial exhaust hood was developed, and a sensitivity of exhaust hood performance as a function of key geometrical parameters was carried out, with the aim of optimizing the pressure recovery coefficient and minimizing the overall dimensions of the exhaust casing. Since hood performance strongly depends on a proper coupling with the turbine rear stage, such a stage was modeled using the so-called mixing-plane approach to couple both stator–rotor and rotor-diffuser interfaces. A detailed analysis of the flow field in the exhaust hood in the different configurations was performed, detecting the swirling structures responsible for the energy dissipation in each simulation, as well as correlating the flow field with the pressure recovery coefficient.