Aggregate interlocking allows transferring shear and normal stresses through open cracks, and is considered to significantly contribute to the force transfer in cracked concrete. The complex phenomenon depends on the roughness of cracked surfaces, where material protruding from one side may engage with the opposite one. Two-Phase models were established in the 1980s by Walraven to estimate the force transfer, distinguishing between cement matrix and spherical aggregates. The approach leads to good results but has several shortcomings. In this paper, the fundamental assumptions are reviewed using specific numerical and experimental investigations. Special tests respecting the geometrical assumptions are presented and the results compared with numerically calculated estimates. The model is extended to address some shortcomings and investigate the physical nature of the main parameters. Positive aspects of Two-Phase models and a number of limitations are highlighted, allowing a consistent step forward in the understanding of aggregate interlocking.