This paper assesses a number of direct strength methods (DSM) for thin-walled steel members under combined major axis bending and compression. These methods consist of two ways of calculating the compression – bending moment (P-M) interaction surface (plastic or first yield) of the cross-section, and the following three methods of operating on the P-M surface: using a compound value of the surface; converting a beam-column to an equivalent beam; and converting a beam-column to an equivalent column. The aim of this research is to identify the most consistently accurate and easy to implement DSM equations to calculate the resistance of thin-walled steel beam-column members, by comparison against numerical simulation results, obtained using a validated ABAQUS model. The numerical simulations investigated the effects of changing different design parameters to ensure a sufficient dataset for each failure mode, including section size, member length, flange restraint, load eccentricity and yield strength. Based on an evaluation of the mean and standard deviation of ratios of the calculated resistances to ABAQUS simulation results, it has been found that for global buckling, any of the methods can produce accurate results. For local buckling, the standard deviation values for all the methods are similar, about 0.05. However, when judging the mean values, the best methods are the equivalent beam method using plastic P-M surface, with a mean value of 0.995. For distortional buckling, upon modification of the existing DSM equations for pure bending using the plastic P-M surface, the equivalent beam method gave an average of 0.998 and a standard deviation of 0.102. Overall, the equivalent beam method using the plastic P-M surface is the most consistently accurate method.