The prediction methods for dispersed flow film boiling (DFFB) are reviewed. The correlation work, based on modifications of the Dittus-Boelter correlation for single-phase flow, is discussed first. The merits and limits of this approach are analysed, and the need for more advanced methods accounting for thermal non-equilibrium is evaluated. The phenomenological models aiming to calculate the vapour superheat are discussed, highlighting the assumptions used in their development. Better predictions are possible by using mechanistic models, where the empiricism is shifted to the closure laws. It is shown, however, that even the latest models do not consider important phenomena such as droplet break-up and liquid fractions cross-sectional distribution, and this failure can cause severe errors in the prediction of the cooling effectiveness of the mixture. Finally, a new model aiming to overcome these limitations and based on the Lagrangian description of the droplet hydrodynamics is briefly outlined.