Shaft loaded blister tests were performed to study the interfacial fracture behaviour of a polymer composite coating material adhesively bonded to a metal substrate. For this purpose, the energy release rate required to propagate the crack along the polymer composite coating–metal interface was measured. To do this, six different initial blister radii (a), three different loading rates and two different spherical ended shaft sizes were employed. Fresh water conditioning at 55°C for 1000h was also carried out on the blister specimens to study the environmental effect on the total energy release rate (GT) of initial blister debonding. For a 5mm diameter shaft, with the increase of loading rate, the GT values were found to increase for the controlled samples while the loading rate followed the similar trend with lower GT values for the conditioned samples. The GT values obtained using two different shaft sizes showed that the larger shaft diameter had higher GT values than the smaller shaft diameter. Conditioned specimens tested using both of the loading shafts were generally found to have lower GT values than the controlled specimens. A comparison with the linear and nonlinear finite element analyses reveals the application ranges of the solutions for determining the energy release rate based on the plate bending model or the nonlinear membrane model. It is shown that the plate based energy release rate solutions should be used for a small blister and the membrane based solution should be adopted for a large blister.