Critical variables that underlie the performance of a national sample of young adults on a diverse set of document literacy tasks were identified. The final sample was 3,618 adults. The identification of these variables provides an important first step toward building a theoretical model that would systematically account for the constructs of document processing. The 61 tasks and their associated documents that make up the document scale of the National Assessment of Educational Progress (NAEP) Young Adult Literacy assessment were parsed using a specially devised grammar. Based on the parsings, variables were identified to account for the probability of success for the total population and for major subgroups of interest. The identified variables accounted for 89% of the variance for the total population. Among racial and ethnic groups, these variables accounted for 89% of the variance for White, 81% for Black, and 87% for Hispanic young adults. Among levels of education, these variables accounted for 56% of the variance for young adults with 0 to 8 years of schooling, 81% for young adults with 9 to 12 years of schooling, 88% for young adults with high-school degrees, and 84% for young adults with post-high-school degrees. Findings are discussed in terms of the need to provide a more general framework for describing, comparing, and researching documents. Five appendixes discuss parsing and scoring. (Contains 2 figures, 3 tables, 1 appendix table, and 70 references.) (Author/SLD)