This paper presents recent research carried out at the University of Sydney on the full-range behaviour of top-and-seat angle connections. A total of 18 tests were conducted, covering two connection configurations and three loading conditions, including bending, combined bending and compression, and combined bending and tension. The highlights of the experimental programme include: a) all connections were tested to their complete failure, thereby investigating the full-range behaviour of the connections, including initial stiffness, ultimate bending strength, ductility and failure mode, as well as the detailed behaviour of the top flange cleat component; and b) multiple tests were carried out on nominally identical connections to examine the variability of joint behaviour, providing valuable information for estimating the uncertainties in joint properties due to the random variation in geometrical and material properties, including imperfections introduced during fabrication. Based on the experimental results, conclusions were drawn on the applicability and accuracy of the Eurocode 3 Part 1-8 provisions for calculating the initial stiffness and ultimate strength of top-and-seat angle connections. Results showed that the Eurocode 3 model significantly underestimates the connection strength by over 80%, due to the conservative strength prediction of the top flange cleat component. To overcome this limitation, a new mechanical model was developed for the top flange cleat component to more accurately predict its full-range behaviour, from initial linearity to complete failure. The model reflects the real deformation pattern of a flange cleat by capturing the development of four plastic hinges in the cleat, while remaining simple and applicable to hand calculation. The model was applied to the tested connections, showing a conservative and reasonably accurate strength prediction with substantial improvement from the current Eurocode 3 model.