To examine whether children's acquisition of perspective-marking language supports development in their ability to reason about mental states, we conducted a longitudinal study testing whether proficiency with complement clauses around age 3 explained variance in false-belief reasoning 6 months later. Forty-five English-speaking 2- and 3-year-olds (23 female, Time 1 age range = 33-41 months) from middle-class families in the North-West of England took part in the study, which addresses a series of uncertainties in previous studies. We avoided the confound of using complement clauses in the false-belief tests, assessed complement-clause proficiency with a new comprehensive test designed to capture gradual development, and controlled for individual differences in executive functioning that could affect both linguistic and sociocognitive performance. Further, we aimed to disentangle the influence of two aspects of complement-clause acquisition: proficiency with the perspective-marking syntactic structure itself and understanding of the specific mental verbs used in this syntactic structure. To investigate direction of causality, we also tested whether early false-belief reasoning predicted later complement-clause proficiency. The results provide strong support for the hypothesis that complement-clause acquisition promotes development in false-belief reasoning. Proficiency with the general structure of complement-clause constructions and understanding of the specific mental verbs "think" and "know" in third-person complements at Time 1 both contributed uniquely to predicting false-belief performance at Time 2. However, false-belief performance at Time 1 also contributed uniquely to predicting complement-clause proficiency at Time 2. Together, these results indicate a bidirectional relationship between linguistic and sociocognitive development.