A discussion of second language program evaluation focuses on the interpretability of test scores as a criterion in program evaluation. It looks at both test design and research design issues. First, eight method-comparison, program evaluation studies that compare acquisition-based and analysis/practice based methods are described. Acquisition methods are defined as those in which students are exposed to the language as a whole, anticipating that the student will acquire structure subconsciously; traditional methods also use analysis, practice, and explanation to build language competence. Then two test-design issues are examined: test content (achievement vs. proficiency) and the kind of scale (norm-referenced vs. criterion-referenced) used. The choices made in the eight studies are analyzed. Next, the effect of research design on interpretation of test scores is examined, with particular attention given to two issues: the purity of the instruction used and the background of the subjects. Two of the studies are selected as more easily interpreted than the others. The results of the studies are then summarized and compared, and implications for test development and valid use of test results in program evaluation are outlined. (MSE)