A model for predicting the overflow rate from a column of rising foam is presented. This model assumes that the Plateau borders have the cross-sectional geometry the-same as the void between three mutually contacting cylinders. Another assumption is that the Plateau border walls are rigid. This theory is tested against experimental data for the overflow rate of a rising foam of water/glycerol mixtures stabilised with sodium dodecyl sulphate. The theory is seen to over-predict the observed overflow rate; this discrepancy is attributed to the fact that the Plateau border walls probably do not exhibit the property of rigidity. An empirical approach to predicting overflow rate that uses a Richardson and Zaki-type expression for the hindered rising of bubbles is seen to have far greater success than the mechanistic model. The applicability of experiments and theories for gas-liquid foams to mineralised froths is discussed. (C) 2003 Elsevier Ltd. All rights reserved.