The generalization that V-to-I movement is conditioned by rich subject agreement on the finite verb (the Rich Agreement Hypothesis) has long been taken to indicate a tight connection between syntax and morphology. Recently, the hypothesis has been questioned on both empirical and theoretical grounds. Here, we demonstrate that the empirical arguments against this hypothesis are incorrect and that it therefore must be rehabilitated in its strongest form. Theoretically, we argue that the correlation between syntax and morphology is not direct (morphology does not drive syntax) but follows from principles of language acquisition: only if language learners are confronted with particular morphological contrasts do they postulate the presence of corresponding formal features that in turn drive syntactic operations.