The present activity was carried out in the framework of the Clean Sky European Research Project ITURB (optimal high-lift turbine blade aeromechanical design), aimed at designing and validating a turbine blade for a geared open-rotor engine. A cold-flow, large-scale, low-speed (LS) rig was built in order to investigate and validate new design criteria, providing reliable and detailed results while containing costs. This paper presents the design of an LS stage and describes a general procedure that allows to scale three-dimensional (3D) blades for LS testing. The design of the stator row was aimed at matching the test-rig inlet conditions and at providing the proper inlet flow field to the blade row. The rotor row was redesigned in order to match the performance of the high-speed (HS) configuration, compensating for both the compressibility effects and different turbine flow paths. The proposed scaling procedure is based on the matching of the 3D blade loading distribution between the real engine environment and the LS facility one, which leads to a comparable behavior of the boundary layer and hence to comparable profile losses. To this end, the datum blade is parameterized, and a neural-network-based methodology is exploited to guide an optimization process based on 3D Reynolds-averaged Navier–Stokes (RANS) computations. The LS stage performance was investigated over a range of Reynolds numbers characteristic of modern low-pressure turbines (LPTs) by using a multi-equation, transition-sensitive, turbulence model. Some comparisons with experimental data available within the project finally proved the effectiveness of the proposed scaling procedure.