In the operation of mechanical flotation cells, the dispersion of gas into fine bubbles may be expressed by three indicators : bubble size, gas holdup and superficial gas velocity. Taken together, these properties determine the bubble surface area flux (S-b) in the cell, which has been found to have a strong correlation with the flotation rate constant (k). Previous work by the authors has indicated that it is possible to predict the value of k for a known are in a cell from a knowledge of the bubble surface area flux generated in that cell. In order to make good use of this finding, an empirical model has been developed to predict S-b in mechanical flotation cells, using data from extensive pilot industrial scale test programs. The model is able to predict S-b from cell operating conditions, impeller design and feed properties. The model has been validated for different types and cell sizes, impeller types and ore types, in different independent investigations carried out at several concentrators in Australia and South Africa. This paper outlines the development of the model, the parameter estimation, and the validation using a number of additional data sets.