Forty-four children (mean 3.8 years) were given three false belief, a working memory, and four language tasks (each designed to tap a different aspect of syntax or semantics), and were tested again 6 months later. Once the range of scores in the language and false belief tasks were equated, there was a bidirectional relation between language and theory of mind. There was no evidence for syntax playing a unique role in the contribution of language to theory of mind. No one measure of syntax or semantics was more likely than any other to predict later false belief. Nor was false belief related more to one aspect of later language (syntax vs. semantics) than another. Our data, taken with other findings, are consistent with the idea that both syntax and semantics contribute to false belief understanding. Working memory did not mediate the relation between language and theory of mind, nor did it facilitate later false belief.