Applications of expert system technology in the urban planning domain range in complexity from the relatively simple task of representing existing manually held knowledge bases within a single shell, to complex exercises in knowledge elicitation and modelling within a multi-system environment. We suggest that the complexity of expert system construction relates to the nature of the knowledge to be contained within the system. In particular, it depends on the degree to which knowledge is preprocessed and on the mix of knowledge types. After reviewing the concepts of knowledge, the knowledge base and knowledge capture with respect to urban planning subject matter, two contrasting systems are examined which illustrate this range of approaches. The first can be thought of as a text animation system since its expertise lies essentially in the ability to navigate a complex document in a way that the professional planner might. The second represents the more protracted exercise of capturing and modelling human expertise and involves the merging of GIS and expert system technologies.