Background: The steel beam-to-column connections are vulnerable structural elements when a building loses one or more of its vertical load-carrying components due to abnormal or accidental loading conditions. After a column is destroyed by abnormal loads, the tensile axial force of the beam gradually increased, while the bending moment decreased, and the load-resistance mechanism shifts from a flexural mechanism to a catenary mechanism, with the axial force becoming the prevailing factor. Aims: This paper investigates the progressive collapse performance of steel beam-to-column connections. While undergoing large deformation, the beam-to-column connections are subjected to moment, shear, and tension in conjunction with high ductility demand. Their behavior under monotonic loading depends on the moment-axial tension interaction and greatly affects the progressive collapse resistance of the structure. This paper presents a critical review of experimental tests of different types of steel beam-column joints (flexible, rigid, and semi-rigid) under a central-column-removal scenario. Methods: The experimental results, including load-deformation relationships, failure modes, and catenary effects, are described in detail. The findings are used to evaluate the rotation capacity of different types of steel beam-to-column connections. The results are compared to the acceptance criteria specified by the main progressive collapse guidelines for several beam-to-column connection categories. Results: In simple (flexible) joints, the stiffness and strength at higher drift angles essentially depend on the tensile capacity of the connection that prevents, in some cases, the full development of the catenary mechanism. The connection depth alone does not seem to be an effective parameter to predict the rotational capacity of beam-to-column connections, since different connections with similar values of the connection depth result in very different values of the maximum rotation capacity. In fully rigid and semi-rigid connections, after the column removal, the flexural resistance controls the behavior at the preliminary phase, and the tensile force is almost zero. With increased downward displacement, the axial tensile force also increases, developing a catenary mechanism. Although the stiffness of rigid and semi-rigid connections is higher than flexible connections, both categories result in similar rotation capacity. Conclusion: In all the simple connections herein considered, the plastic rotation capacity obtained by tests was found much higher than the code recommended values that are probably too conservative. On the contrary, for one rigid and two semi-rigid connections, the values of the plastic rotation capacity obtained by tests are lower than the corresponding recommended values. Thus, the suggested acceptance criteria proved to be out of the conservative side.