This analysis reveals critical weaknesses surrounding the theoretical basis for the AIR (American Institutes for Research) evaluation design. It poses questions concerning the evaluation methodology. Specifically, it identifies major discrepancies in the identification of the target population, the selection of comparable control groups, test instruments used in the study, the amount of time between pre and post testing, lack of consistency in programs being studied, adequacy of instructional staff, and the source of funds being used. The major issues discussed include: (1) identification of language groups, (2) comparability of treatment and comparison groups, (3) instrumentation, (4) pre and post testing, (5) variability of instructional treatment, (6) variability of teacher characteristics, and (7) costing of bilingual education. It is the opinion of IDRA that inadequacies in methodology, sampling, instrumentation, controls, and implementation make the AIR findings highly questionable. The information gathered by AIR is not entirely useless, however, since it is obvious that there are problems and discrepancies in the implementation of bilingual education programs. It is recommended that the U.S. Office of Education should diligently pursue further analysis into these problems, and channel federal resources toward their solution. (Author/AM)