The present study examined the unique contributions of readers' goal and text structure on comprehension. In Experiment 1, participants read procedural and descriptive passages to perform the procedures, summarize the passages, or to answer questions. The perform goal showed highest comprehension, with no difference due to text type. In Experiment 2, participants read either a narrative-like or a list-like procedural text. The narrative-like text led to better comprehension than the list-like text for the perform condition, but this finding was reversed for the other conditions. The results indicate that past differences on text type were based on differences on local coherence.