S. E. Svensson recently argued that the lateral-torsional critical buckling stress sigma //e//l for a beam with one flange rigidly restrained along its entire length could be obtained by solving the approximately equivalent buckling load problem obtained by treating the free flange as an elastically supported column and dividing by the flange area. The present paper suggests that it would be safer to add 15% of the web area to the flange area when finding sigma //e//l, on the basis of a comparison with accurate lateral-torsional critical buckling stresses which the exact thin plate theory computer program VIPASA gave for pure bending of four representatively chosen beam cross sections. Svensson's design curves and table for the equivalent elastically supported column covered several practical variations of the axial force along the length, but were restricted to simply supported ends. The present paper gives design curves which are much more comprehensive.