In frame structures, consisting of columns and beams, gravity loading on beams (floors) leads to moments as well as axial forces in the columns. In cases with slender columns, maximum moments may develop between columns ends due to second order effects. Approximate methods for calculating maximum moments between column ends are given in Norwegian and international standards. These ``conventional'' methods are formulated in terms of approximate effective lengths and a moment gradient factor (often labelled Cm), and are mainly developed for structures of regular geometry, and in which the beams and neighbouring column(s), that frame into an end of the considered column, provide a positive rotational restraint at that column end. In some circumstances, the provided rotational end restraint may actually become negative. The main object of the present work is to study the applicability of conventional methods to such cases, and other more commonly encountered cases. For this purpose, approximate maximum moment predictions are compared to results of exact, linearized second order calculations for a wide range of parameters. Similar comparisons are also made for some other approximate methods having alternative moment gradient factor definitions and differently computed effective lengths.