In two experiments, we investigated the relationship between reading ability and linguistic knowledge in adults. The results from Experiment 1 showed that good comprehenders performed more accurately than average comprehenders in a syntactic-judgment task that required them to decide whether pairs of words served the same grammatical function in sentences. By contrast, the reader groups performed similarly when required to make semantic judgments about whether pairs of words were related in meaning. In Experiment 2, individuals were classified according to comprehension level and reading speed. Good comprehenders again performed more accurately than average comprehenders in the syntactic task but not the semantic task. We argued that differences in form-class knowledge could be associated with corresponding differences in syntacticprocessing efficiency, and thus with variation in reading-comprehension skill generally.