In this paper we examine the operationality/generality trade-off and how it affects performance of explanation-based learning systems. Experience with the ARMS learning apprentice system, presented in the form of an empirical performance analysis, illustrates both sides of the trade-off.