In spite of considerable amount of research on flat plate structures, understanding the brittle and catastrophic punching shear failure is still somewhat incomplete. The thesis focuses on the punching shear behaviour of interior flat slab-column connections under gravity loading. A Comprehensive literature review has shown that although there are several different mechanical models available to calculate punching shear capacity, they are not only complex and difficult to apply in engineering practice but also are not precise. Code provisions such as ACI 318-08 (2008), Eurocode 2-2004, CEB-FIP Model Code 1990 and BS 8118-1997 are all based on empirical equations which were developed using limited experiments from the literature. Therefore, the precision of punching shear capacity prediction using the code equations are questionable. Previous researches assessed the accuracy of the code provisions only by comparing with their limited set of experiments. In this thesis, the author has constructed a large database using 549 punching shear experiments from the literature and concluded that to assess the code provisions in a much more rigorous manner, it is essential to evaluate the effect of the key parameters that affect the punching shear capacity of flat slab-column connections using this database. This study has shown that BS 8110 (1997) predicts punching shear capacity with high accuracy while ACI 318-08 (2008) underestimates the punching shear capacity for slabs with low flexural reinforcement. Lenton Steel Fortress (LSF) type of shear reinforcement has a better anchorage behaviour compared to stirrups. However, experiments with this type of shear reinforcement are very limited. In order to gain an insight of the behaviour of slab-column connections under gravity loading, two series of experiments were conducted in this study: 72 pullout tests using LSF strips and 3 real scale slab-column connections. LSF strips have been shown to have enhanced anchorage behaviour and the use of LSF type shear reinforcement has resulted in an increase of 67% punching shear capacity and 152% in deformation capacity. In addition to the experiments, numerical modelling was carried out to further investigate the behaviour of flat slab-column connections. It is also concluded that Finite Element analyses using ABAQUS is capable of predicting the behaviour of such connections with sufficient accuracy. Using the validated numerical models, a parametric study was carried out to investigate the effect of parameters such as column dimensions, slab depths, top and bottom reinforcement ratios and shear reinforcement area on the punching shear capacity of the flat slab-column connections. A development of a simplified shear model was achieved by Bayesian Neural Network (NN) using the parameters previously determined from the comprehensive databases and numerical analyses. The simplified shear model that was developed by the author predicts punching shear capacity with high accuracy.