The gas-perturbed liquid model of Zhang et al. (1995) is modified in an effort to improve its prediction of the minimum liquid velocity of fluidization, Ulmf, of a bed of solid particles in the presence of a low or moderate cocurrent flow of gas. Variants of the model are considered where the buoyancy term is based on the gas-liquid mixture, instead of the liquid alone, and with the frictional pressure gradient given by several alternative equations to the Ergun equation employed in the original gas-perturbed liquid model. All versions of the model provide similar dependence on such factors as gas velocity, particle diameter, particle density and liquid viscosity as those seen experimentally. The mixture buoyed equation with the drag based on an equation suggested by Foscolo et al. (1983) gives improved predictions over the original Zhang et al. (1995) model, but the best overall agreement is with buoyancy based on the liquid alone and the first term in the drag equation with the Carman (1937) constant of 180 instead of Ergun's 150. The predictions are sensitive to the minimum fluidization voidage, which is measured, assumed, or estimated. Further work is required to investigate minimum liquid fluidization velocities experimentally for particles of density closer to that of the liquid, and for high-viscosity liquids.