Portfolio assessment was investigated as a method of evaluating elementary school students' competence in writing. The study comprised an empirical study of the utility and meaningfulness of using a holistic/analytic scoring rubric to score students' portfolios and a qualitative analysis of scoring approaches drawing on raters' critiques of the analytic scoring method. The site was an elementary school participating in the Apple Classrooms of Tomorrow project. Portfolios were composed of a working file and a smaller showcase file of students' best pieces (student-selected). The scoring rubric was a well-validated system with subscales reflecting key aspects of competent writing. Interrater agreement was determined for three raters evaluating standard writing assessments, students' portfolios, and classroom work in sets of writing unidentified by student for grades 1, 3, and 4. Thirty-five portfolios were evaluated. Results demonstrate that holistic rating of class work and portfolio collections can achieve high interrater agreement and can discriminate among grade levels and genre differences. Findings also demonstrate the complexity of developing large-scale portfolio assessments. The paper includes 16 tables of study data and a 15-item list of references. An appendix contains writing assignment directions. (SLD)