Previous research shows that high-functioning children with autism are slow to pass "litmus" false belief tests of ToM but how this may relate to other aspects of mindreading (e.g., discerning thoughts from facial expressions) is less clear, partly for methodological reasons. Thus the joint methodological and conceptual goals of this study were: (1) to devise and psychometrically validate a new, simplified eye-reading test for preliterate children with or without autism and (2) to use the new test to explore links of false belief understanding with eye-reading in children with autism and matched control groups. A false belief battery and the new eye-reading test were given to 87 Australians: 22 children with autism aged 6-13 and 65 typical developers in three control groups (11 age-matched primary-schoolers; 37 ToM-matched preschoolers and 17 adults). Results supported the new test's psychometric validity and showed that, for children both with and without autism, false belief and eye-reading were significantly correlated. A hierarchical multiple regression showed this association was independent of age, gender and diagnosis. Although adults earned higher eye-reading scores overall, children equalled them on 44% of items. Implications of the findings for future use of the new test, and for explanations and interventions on behalf of ToM development in autism, were considered. (Contains 2 tables.)