An approach to simulate the performance of an active noise control system is proposed. The simulated result is shown to be a good approximation to that of a real system. By means of this approach some examples which may not be easily realisable in practical implementations are studied. Results concerning causality and filtered-x-type algorithms are interesting. Simulations show that the investigated active noise control system needs a considerable causality margin to perform well. This is explained by the inverse modelling nature of the selected feedforward controller. An example of the delayed inverse modelling of a nonminimum-phase system is given in which the quality of the inverse modelling depends on the delays.