Hydrometallurgical processes are difficult to describe fundamentally, owing to their largely stochastic nature and the often ill-defined chemorheology of the froth. Although these processes are consequently difficult to monitor accurately by means of classical methods, progress has recently been made with regard to the use of neural net control systems. In this paper the use of a self-organizing neural net to monitor the behaviour of an industrial platinum flotation plant is discussed. The net is shown to be an efficient means of detecting arbitrary small changes in process conditions. In addition to the self-organizing neural net, the performance of a fuzzy ARTMAP system is also evaluated. These types of nets are capable of robust incremental assimilation. of new process knowledge, as is demonstrated in terms of the classification of flow regimes in an air-water flow system.