In this discussion of the use of achievement tests in evaluating Title I programs, matching content between test and curriculum is the main theme. The point is made that unless a test measures what is taught, it cannot be sensitive to whatever gains the instruction produces. Thus, if different instructional treatments have different objectives, it is unlikely that a single test will provide equally valid assessments of their effects. Where different tests are needed to assess the effects of different curricula, national norms can provide a common metric for comparing gains measured with different tests. A gain of one-third of a national standard deviation on Test X, for example, would be considered equal in value to a gain of the same size on Test Y assuming that the two tests were equally good matches to the corresponding curricula. This approach will work quite well, presumably, when "typical" standardized reading or mathematics achievement tests are used. It breaks down if both test and curriculum are very narrowly focused. Under these circumstances gains will appear too large. (Author/CTM)