Froth flotation is one of the most widely used beneficiation processes for fine and ultrafine coals and minerals. Maintaining optimal metallurgical performance in industrial flotation operation is still a challenging task owing to the lack of an effective tool to diagnose the flotation performance in real_x0002_time. For this purpose, a number of flotation performance monitoring systems with different working principles have been developed to provide instant feedback for process control and optimisation. In this paper, for the first time, five different flotation performance monitoring systems were systematically compared through an on-site trial in a coking coal flotation plant in Queensland, Australia. The monitoring systems were used to measure (i) gas holdup, (ii) overflowing froth velocity, (iii) overflowing froth height, (iv) air recovery and (v) drag force exerted by overflowing froth, in an industrial flotation column and the outputs of these systems were collected simultaneously at a wide range of operating conditions. For each monitoring system, the best regression model describing the correlation between the system output and the flotation performance was selected from various regression models (eg linear, quadratic, power-law and logarithmic) using the Akaike information criterion (AIC). A comparison of the coefficient of determination (R2 ), residual sum of squares (RSS) and measuring accuracy at 95 per cent confidence level of the best regression models of these monitoring systems suggested that the drag sensor outperformed the other systems at a wide range of flotation operating conditions. The superior performance of the drag sensor was attributed to its distinctive design that enabled intimate sensing of variation of the overflowing froth properties. Furthermore, the drag sensor had the ability to monitor feed ash content variation when the other operation variables were set almost constant.