The effectiveness of externally bonded (EB) fibre-reinforced polymer (FRP) composite reinforcements depends on their bond behaviour with the specific substrate onto which they are applied. Direct shear tests are commonly used to obtain experimental results on the bond behaviour of carbon FRP (CFRP)-concrete joints. These results are often modelled using analytical and numerical models that assume a fracture mechanics pure Mode-II loading condition and a zero-thickness CFRP-concrete interface with cohesive behaviour. Numerous models proposed in the literature showed a load response snap-back phenomenon, which is usually not captured experimentally. This paper provides and discusses the experimental results obtained from single-lap direct shear tests of CFRP-concrete joints. The tests are conducted to capture the load response snap-back by accurately controlling the CFRP-concrete interface crack propagation. The results obtained are modelled using an analytical approach based on a rigid-bilinear cohesive material law (CML) that describes the CFRP-concrete interface behaviour. A comparison between experimental and analytical load responses is provided. The approach adopted is proven effective in accurately capturing the CFRP full-range bond behaviour, including the snap-back branch.