A study was conducted to refine the writing scale incorporated into the Peabody Individual Achievement Test-Revised. The test uses a single scale for judging writing samples from students in grades 2 through 12. It was questioned whether a single, relatively brief rating scale could have the sensitivity required to discriminate among the range of written expression presented by students across this broad grade range. The written expression subtest provides the students with 20 minutes to write a story in response to one of two picture prompts. Each of 24 discrete items is rated on either a dichotomous or trichotomous scale, depending on the characteristic being rated. Each item on the scale yields a point score. For this study, 1,087 writing samples were collected from students in grades 2 through 12 in the spring of 1986. About half of the students took Prompt A, and the other half took Prompt B. A subset of 168 students was selected to retake the alternate prompt for later use in estimating alternate-forms reliability. Item weighting methods applied to the data include regression, graded-response, and teachers' rating methods. Ten secondary school English teachers rated the 24 written expression items in terms of their importance to the evaluation of a student's written composition. Application of these three methods failed to yield improvements either in mean raw score growth from grade-to-grade or in internal consistency reliability. Five data tables are presented. (TJH)