Immune systems of life forms have been a rich source of inspiration to computer scientists. Problem solving strategies, borrowing ideas from the immune system analogy, have been successfully applied to challenging problems of modern computing. However, no formal model of a typical immune system has so far been proposed. This paper reports on an experiment towards modelling an immune system using process algebra. Our model of the immune system separates individual components. Each element is specified in terms of its observable behaviour. This paper also provides a formal specifiation of some properties of an immune system, including robustness and self-stabilisation. It outlines the sort of mechanisms that are required to formally verify the resulting model.