Since the early days of flotation, process technicians have known that the speed at which the froth is recovered over the lip of the cell has a very direct and consistent influence on the grade and recovery of the circuit. No plant would deny that, even to this day, one of the most important aspects of a process technician's job is ensuring that the froth is moving at a desired and controlled speed over the lip. Unfortunately, process technicians cannot watch the froth throughout the entire shift, as there are always other problems to attend to. A considerable amount of work has thus been done to develop a system' that automatically measures the speed of the froth. This measurement would then be used in simple controllers to emulate the steps taken by the process technician in correcting deviations of the froth speed from its desired level. Many of the earlier attempts to measure speed were ineffective, as they were unable to cope with turbulence on the froth surface. The FrothMaster instrument makes use of machine vision technologies to measure the speed of the froth over the lip of a cell at a very high sampling rate. This instrument produces an extremely reliable and accurate measurement of the froth speed (as well as other parameters such as bubble size and froth stability). As we will show, the effectiveness of the speed measurement in controlling the performance of the plant is a testimony to its reliability. Three FrothMaster units have recently been installed on the Cadia Hills Gold Mine in New South Wales (Australia). The objectives of this installation were:to control the mass-recovery of two Tank Cells in series on the same level where differences in the hydrostatic head between the two cells result in the cells pulling inconsistently with respect to each other;control of the concentrate grade (per cent copper) from the first rougher;reduction in the frother consumption in the rougher circuit by monitoring the froth speed. A simple controller was designed to react to disturbances in the same way that human operators react. This was done after consultation with the metallurgists and the process technicians at Cadia. The controller had two functions: Stabilisation: The stabilising controller made use of three manipulated variables (level, frother addition rate and aeration rate) to control the froth speed to a setpoint. Optimisation: The optimising controller adjusted the froth speed setpoint based on deviations of the grade (measured in real-time by the OSA) from a setpoint. The plant metallurgist adjusted the grade setpoint based on information about the ore. The results from a 21-day trial have shown the following:The ratio of concentrate mass-recovery between two cells in series on the same level has been successfully balanced, thereby overcoming problems due to differences in hydrostatic heads.It is possible to control the speed of the froth, in mm/s, to within five per cent of its setpoint using the aforementioned manipulated variables. It has also been shown that, for reliable control, it is necessary to control more than just one variable (eg level or air).The grade of the first rougher concentrate has been controlled to within 0.3 per cent of the grade setpoint for long periods of time, which has important implications for the circuit recovery.The frother consumption has been reduced significantly. This paper will thus show that substantial progress has been made to demonstrate the value of using on-line froth speed measurements in flotation plant control to obtain real financial benefits.