Reduced beam sections (RBS) are increasingly used in modern construction due to their large rotational capacity and ability to dissipate and absorb large amounts of seismic energy, thus, creating a ductile and stable steel frame system. Currently, in the design of RBS connections, the effect of RBS cutting parameters on the cyclic performance of the beam elements are not taken into account. However, using different RBS geometries for any single beam, compared to its full section, can have up to 30% differences in cyclic behaviour of the connections. The aim of this study is to develop a more efficient design methodology for RBS connections, by investigating the cyclic performance of different beams with a wide range of different flange reductions. First, detailed Finite Element (FE) models of different American Wide Flange RBS connections are developed and validated against two cyclic beam-column sub assembly experiments from literature. The models took into account the non-linear material properties and adopted appropriate modelling techniques for the connection welds, supports and bracing. Then, an extensive parametric analysis on 90 different specimens was undertaken in order to assess how the geometrical parameters which define RBS connections affect the key design parameters including, Yield Moment (My), Peak Moment (Mc), Ultimate Rotation (θu), Ductility (μ) and Energy Dissipated (Ediss). It is shown that the depth (“c”) and width (“b”) of the RBS cut are the most influential geometrical design parameters, affecting up to 30% changes in the key performance parameters compared to a full beams section (no RBS present). Finally, based on the results of this study, practical design equations are proposed to predict the seismic performance of RBS connections compared to a full section (no RBS present) as a function of the five key design parameters used in common practice. The proposed equations should prove useful in preliminary design of RBS connections to achieve maximum seismic performance.