A case is presented for adjusting the scores for free response items in the Advanced Placement (AP) tests. Using information about the rating process from the reliability studies, administrations of the AP test for three subject areas, psychology, computer science, and English language and composition, are analyzed. In the reliability studies, 299 psychology essays, 250 and 248 essays for two forms of the computer science examination, and 250 essays for English were rated. It is shown that the minimum squared error score adjustments proposed by N. T. Longford (1993) result in changed AP grades for an appreciable percentage of examinees. The proposed schemes are easy to implement and involve no iterative procedures. Four tables and two figures present details of the analyses. (SLD)