Despite having been relegated to the realm of superstition during the dominant years of behaviourism, the investigation and discussion of consciousness has again become scientifically defensible. However, attempts at describing animal consciousness continue to be criticised for lacking independent criteria that identify the presence or absence of the phenomenon. Over one hundred years ago William James recognised that mental traits are subject to the same evolutionary processes as are physical characteristics and must therefore be represented in differing levels of complexity throughout the animal kingdom. James' proposals with regard to animal consciousness are outlined and followed by a discussion of three classes of animal consciousness derived from empirical research. These classes are presented to defend both James' proposals and the position that a theory of animal consciousness can be scientifically supported. It is argued that by using particular behavioural expressions to index consciousness and by providing empirical tests by which to elicit these behavioural expressions, a scientifically defensible theory of animal consciousness can be developed.