Internal blade cooling cavities show ever increasing geometrical complexity. Consequently, a deep insight on the flow behavior is mandatory for a reliable design, while trustworthy experimental databases are definitely needed to assess the reliability of numerical prediction tools. Present contribution provides a detailed investigation of the flow field inside a modern internal cooling passage for turbine blade trailing edge. The investigation is carried out by coupling the results of accurate Particle Image Velocimetry (PIV) measurements and steady state CFD predictions. The channel is characterized by an high aspect-ratio trapezoidal cross section of width and height that reduce progressively moving from blade hub to tip and trailing edge, respectively. The flow has a mixed radial-axial direction, it enters the channel at the blade hub, and it is discharged radially at the tip and along the trailing edge. In this latter exhaust section, 7 pedestals guarantee structural resistance and enhance the turbulent heat transfer. The investigation is carried out at an engine representative Reynolds number of 20000 and for two values of the mass flow split between tip and trailing edge exhaust sections. 2D-PIV measurements were performed on several flow planes, carefully selected to allow a comprehensive reconstruction of the 3D flow field inside the passage. Three main aspects have been considered. The first one was the characterization of the flow field at the inlet, which represents the boundary condition for the numerical problem. The second objective was the definition of the mass flow distribution along the trailing edge slot and at the blade tip. Finally, highly spatially resolved measurements were performed inside the inter-pedestals passages. 3D flow separations and horse-shoe vortex branches were observed on the downstream and upstream pedestals’ surfaces, respectively. In order to verify the accuracy of typical industrial CFD, a comparison of experimental flow field with numerical results obtained with both a commercial (Ansys ® CFX ® ) and an in-house developed (open source toolbox OpenFOAM ® ) 3D RANS solver was performed. The accuracy of the commonly used k–ω SST turbulence model to predict the extent of recirculating flows was investigated and discussed. Both codes provided satisfactory results showing also good confidence in the prediction of the complex 3D flow structures inside the inter pedestal passages.