The paper reports results of the practical application of the self-organising controller to the simultaneous control of two interacting joints of a robot. The rationale for its use and the architecture of this heuristic fuzzy rule-based dynamic control approach are outlined. This shows it to be a relatively simple intelligent knowledge-based system capable of both learning and displaying a variable nonlinear input/output relationship. In both step response and tracking experiments the self-organising controller was found to perform better than a conventional PID controller under similar circumstances, although some tuning was required for the former system.