In the paper a parametric study on the cyclic behaviour of T-stub assemblies is presented. The numerical simulation was performed by using a sophisticated FE model. The applied numerical model has been developed by the authors especially for the cyclic analysis of end-plate type joints, including steel-to-steel and steel-to-concrete joints.The primary aim of the parametric study is to examine the behaviour itself, and to draw qualitative conclusions. Another important aim is to test the developed numerical model capabilities. The main focus of the study is on the end-plate and the bolt. Thus, the changing parameter is the bolt resistance relative to end-plate resistance.The actual values of the parameters are chosen so as to cover all the practical cases, including extremely strong and weak bolts. Here, four characteristic cases are presented. In the case of TS1 ( strong bolt ) the bolts are strong enough to remain elastic during the whole loading process. TS4 represents the other extreme case, when the end-plate is strong enough to avoid considerable plastic deformations ( weak bolt ). Between the two extreme cases, TS2 and TS3 correspond to the case when plastic yielding occurs in both the bolts and end-plate ( medium bolt ).Figure 1 presents the established force-displacement diagrams of all the four specimens. F and u denote the applied force and the corresponding displacement measured at the web edge of the T-stub, whilst F e and u e are the force and displacement corresponding to the first yielding.The results of the parametric study show the applicability of the model to the analysis of end-plate-type connections even in case of extreme parameters. The efficiency of the proposed model is justified. The results also well demonstrate the crucial effect of the parameters: the resistance, the rigidity and the energy absorption capacity are strongly affected by the geometrical and material properties of the T-stub assembly. It is also stated that the examined parameters can influence the T-stub behaviour not only quantitatively but also qualitatively.