The relevance of expert systems in agricultural economics research and extension is examined. Some of the limitations of existing expert systems are described. It is observed that expert systems are useful for storing and retrieving knowledge and for solving diagnostic problems but not for mathematical optimisation. The common method of handling uncertainty in expert systems has a dubious theoretical basis. They have no facility for sensitivity analysis. It is concluded that expert systems are unlikely to replace traditional analytical tools for the solution of economic problems but that they may be useful for creation of user-friendly program interfaces.