While school systems most often use achievement test results for individual appraisals, increasing attention to program evaluation and accountability requires that test results be used for institutional appraisals as well. When institutional test results are desired--that is, results for schools or school districts--not all pupils need be tested. Two alternative testing program designs are proposed, for situations where individual test results are desired for some subject areas or grades, and institutional results are desired for others. Seventeen alternative finite-population sampling procedures can be used with these designs. The efficiency of each procedure is evaluated using data from a medium-sized school district, and relative efficiencies are derived. Some sampling procedures are found to yield acceptable precision for estimation of district-mean achievement with a sampling fraction of only five percent. (Author)