This study investigated the effects of imposing task- or process-oriented reading behaviors on reading comprehension assessment performance. Students in grades 5-8 (N = 275) were randomly assigned to hear multiple-choice items read aloud before or after reading a test passage and when they were and were not allowed access to the passage while answering items. A confirmatory factor analysis (CFA) found a one-factor model with all test items loading on the comprehension latent variable fit better than treating literal, inferential, and evaluative items as three separate variables. Subsequently, a structural equation model explored whether testing condition or other covariates (state assessment score and student demographics) explained variance in reading comprehension assessment performance. In grades 5-6, significant positive effects were found for students who kept the text while answering items, regardless of whether or not they previewed the items. In grades 7-8, no significant differences were found for the four testing conditions.