Although there are three main computer-assisted instruction (CAI) formats (drill and practice, simulation, and tutorial), tutorial format is particularly suitable to conceptual school subjects because this format: (1) explains the concept of the text; (2) tests the understanding of the topic; and (3) selects the next step of the subject. When a knowledge module is added to the tutorial format, it is called an intelligent tutoring system (ITS). In short, an intelligent computer-assisted instruction (I-CAI) does emphasize interactions of the learner with the computerized system. In light of methodology and interpretation, this paper examines ITS evaluation studies conducted by leading researchers in the ITS field. Although ITS evaluations greatly influence interest in and support for future ITS work, valid evaluation methodologies are lacking mainly because the system itself is fairly new. The critical component in ITS is its knowledge component, which may not have been evaluated adequately to verify the knowledge. ITS researchers, therefore, have simply suggested using procedures used in evaluation studies of artificial intelligence or expert systems. Such procedures, however, may or may not be applicable in educational settings, and therefore, further research is strongly recommended. Summative evaluations were found to be more difficult to execute than formative evaluations because summative evaluations would involve the comparison of ITS with human tutors using traditional teaching methods across extensive problem domains. (Contains 18 references.) (Author/SLD)