•An effect of subsoil modelling on the test results in the case of footings subjected to punching.•The accuracy of the current and 2nd generation EC2 models for predicting punching capacity of the footings.•An effect of shear slenderness on punching shear capacity of the footings.•Post-punching resistance of the footings resting on the gravel cushion. The results of an experimental program on the punching-shear capacity of foundation footings without shear reinforcement are presented and discussed in this paper, in the case of four tests on as many footings resting on a quasi-natural subsoil (top thin sand layer and bottom thick gravel layer). The design models of EC2 (2004) and prEC2 (2020) for the prediction of the punching-shear capacity are checked as well, by considering 90 tests well documented in the literature and the four tests performed in this research project. Special attention is devoted to the role played by the subsoil of the footing. The largest differences between the test results and the predictions occur in the case of quasi-natural subsoils, while the footings resting on point supports yield the smallest differences. The comparison of experimental and predicted capacities shows that the current model in EC2 (2004) is neither reliable nor safe, for any type of subsoil, since the ratio between the experimental and the assessed values of the load-bearing capacity ranges from 0.625 and 0.810 in 95% of the cases, depending on subsoil properties. On the contrary, the second-generation model introduced in prEC2 (2020) provides better results, since the previous ratio is close to 0.95 in 95% of the cases examined in the paper.