A study of microcomputer courseware reviews was carried out in order to discover which criteria they discuss, whether reviewers agree in their assessments, how evaluative the reviews are, and whether the reviewing policy of the journal influences the review. The content analysis instrument was a composite of criteria developed by organizations that evaluate courseware. A total of 53 variables were included within the following classification scheme: (1) program description and summary; (2) technical aspects; (3) subject matter content; (4) instructional design; (5) presentation; (6) student use; (7) teacher use; (8) review process; (9) evaluation summary; and (10) other miscellaneous considerations such as support materials and field testing. Reviews of ten widely-reviewed logic/problem-solving courseware packages were chosen for study, and content analysis was performed on 147 reviews. Data obtained show that few of the evaluation criteria are used in published reviews and that these reviews are largely descriptive. When reviews do evaluate they are positive and uncritical in nature. It is concluded that courseware reviews need to be improved and that the language of evaluation needs to be standardized. Definitions of evaluation criteria and a list of 25 references are appended. (MES)