This thesis describes research work undertaken to examine the influence of joint resis- tance to in-plane moments on the assessment of five fun scale two dimensional steel frames. All the available information has been collected into a form which is readily usable by the wider research community. The findings have been used to develop improved semi-rigid design methods. Details of the experimental results of five frames are discussed and reported. A variety of joint types were tested, each type of joint exhibited some degree of moment and rota- tional stiffness, leading to frame responses involving significant interaction between beams and columns. This affects both beam and column deflections, the pattern of frame mo- ments, ultimate capacity of members and collapse modes. The principal objective of the tests was to provide experimental data against which two in house sophisticated computer analysis programs may be verified. Details of modifications to an existing databank of the moment-rotation test results to provide a more efficient and user friendly tool for research workers are explained. The main objective of the work described herein was to investigate the influence of semi-rigid connection behaviour on a wide range of subassemblage configurations. A new method for the non-sway column design is developed and validated using the test and analytical results. The new approach is found to be more economical than the existing BS 5950 method but it still essentially conservative. The methods to design the lateral restraint beam including the serviceability and ultimate limit states are also proposed and examined. The most important feature is that the precise form of the M-Ø responses are not important for the semi-rigid design. Finally, a complete set of design methods is advised to take into account the inherent strength and stiffness of semi-rigid joint.