This is second in a series of papers concerned with the performance of a novel technology, the Reflux Flotation Cell. Part I examined the system hydrodynamics, commencing with a gas–liquid system and examination of the fluidization boundary condition. The desliming, or potential to reject entrained fine gangue particles from the product overflow, was investigated by introducing hydrophilic particles. In Part II, a model feed consisting of hydrophobic coal particles and hydrophilic silica was introduced. The separation of these two components was investigated across an extreme range in the applied gas and wash water fluxes, well beyond the usual limits of conventional flotation. The Reflux Flotation Cell challenges conventional flotation cell design and operation in three ways. Firstly, the upper free-surface of the flotation cell is enclosed by a fluidized bed distributor in order to fluidize the system in a downwards configuration, counter-current to the direction of the rising air bubbles. Secondly, a system of inclined channels is located below the vertical section of the cell, providing a foundation for increasing bubble–liquid segregation rates. Thirdly, the system is operated with a bubbly zone, hence in the absence of a froth zone. This combination of conditions provides for the establishment of a high volume fraction of bubbles in the bubbly zone, of high permeability, ideal for promoting enhanced counter-current washing of the rising bubbles, and hence high quality desliming. The arrangement permitted operation at extreme levels in the value of the fluidization (wash water) flux and the gas flux, with the fluidization flux set at up to 2.1 cm/s and the gas flux set at up to 4.7 cm/s for a mean bubble size, db, of 1.5 mm. These gas and wash water fluxes corresponded to a bubble surface flux of 188 m2/m2 s and a positive bias flux of 1.7 cm/s. Thus the operating regime was shown to be far broader than that achieved by conventional flotation, thereby confirming the robust nature of the system. The model flotation feed provided a basis for establishing the flotation performance across this vast regime of operation. Full combustible recovery of fine coal and full rejection of mineral matter were achieved, with good agreement with the Tree Flotation curve. At extreme levels of wash water addition it was possible to selectively strip poorer floating coal particles from the bubble surface, and in turn achieve beneficiation results significantly better than those defined by the Tree Flotation method.