Our ability to infer others’ mental states (e.g., perspectives and beliefs) is integral to successful social interaction. A long-standing debate is whether mentalising relies upon specialised social-cognitive processes or more domain-general systems that coordinate our thoughts and behaviour across social and non-social contexts. In this preregistered study, we investigated this directly by determining the causal links between mentalising and three supposedly distinct domain-general executive functions (EFs): interference resolution (the suppression of task-irrelevant information), set shifting (the flexible switching between cognitive states) and memory updating (using new information to update memory representations). A large sample (N=369) completed an online battery of three experimental tasks that measured each of the EFs and a fourth that measured mentalising – our ability to infer how the world appears from someone’s else viewpoint (level-2 visuospatial perspective-taking; VPT-2) and acknowledge that others’ belief about the world may align with or diverge from our privileged knowledge of reality (true- and false-belief reasoning). Two structural equation models were applied to data acquired with this battery to evaluate direct and indirect pathways between EFs and mentalising. Contrary to our hypotheses, these models suggest that better memory updating and interference resolution increased the cognitive costs associated with VPT-2 and false-belief reasoning, respectively, while set shifting showed no significant effects on mentalising. Furthermore, VPT-2 did not mediate any of the causal pathways between EFs and belief reasoning. This study indicates that EFs contribute to mentalising in subtle and complex ways, and our experimental rigour reveals several methodological considerations for future research.