A capacity theory of comprehension (M.A. Just & P.A. Carpenter, 1992) has provided an integrated account of several central aspects of sentence comprehension, such as the processing of syntactic ambiguity, complex embeddings, syntactic (non) modularity, and individual differences, in terms of the working-memory capacity for language. Some of the evidence supporting the theory is questioned by G.S. Waters and D. Caplan (1996a). This article identifies some of Waters and Caplan's errors about the empirical support in Just and Carpenter (1992), evaluates Waters and Caplan's alternative hypothesis, and presents the results of a new neuroimaging study that supports capacity theory and not Waters and Caplan's separate resources hypothesis.