The effects of prewriting, drafting, and revising in a large-scale writing assessment were studied, focusing on drafting and revision patterns exhibited by students during testing. A total of 465 students were selected for this study (113 in grade 3, 117 in grade 5, and 116 in grade 10). Each student's work included the prewriting, first draft, and second draft of the 1998 Delaware Writing Assessment. The review panel for this study consisted of 10 experienced writing specialists. Of the 475 essays, 34% of the second drafts across the grades showed some evidence of improvement through revisions, but 66% showed no evidence of improvement. Only 3% of the second drafts were considered "greatly improved." Overall, the primary changes students made were on the lexical level of the eight revision categories. A large percentage of students in the lower grades made more changes to the physical appearance and surface level of their writing than did students in the higher grades, but students in the higher grades were more likely to make internal changes. Results indicate that improvement of essays is a function of the total number of changes at each revision level, radical and internal changes, the total change score, and using a prewriting strategy to plan the first draft. Three appendixes contain the scoring rubric and the writing prompts for the assessment. (Contains 27 tables and 17 references.) (SLD)