Column plate numbers, N, were measured for 12 different proteins as a function of mobile phase flow-rate in two gel filtration systems (either denaturing or non-denaturing conditions). These data were used to extend a previous model that predicts bandwidths in reversed-phase and ion-exchange chromatography. Restriction of diffusion of large molecules within column packing pores is now defined more precisely, with a single relationship describing this effect for both reversed-phase and size-exclusion chromatography (SEC) (and presumably other high-performance liquid chromatography systems). Separations by gel filtration (SEC) are now included in our general model. A total of 17 flow-rate studies were carried out, involving different proteins, columns and/or mobile phase conditions (denaturing or non-denaturing). Comparisons of plate numbers predicted by the model with experimental values were satisfactory in 15 out of 17 cases. The remaining two cases appear to represent "non-well-behaved" systems, where experimental bandwidths were higher than predicted values by more than 20%. Initial attempts at understanding the origin of these non-ideal effects are described.