This paper examines the claim that machine induction can alleviate the current knowledge engineering bottleneck in expert system construction. It presents a case study of the rule induction software tool known as Expert-Ease and proposes a set of criteria which might guide the selection of appropriate domains.