Anxiety has been associated with social communication and interaction differences among autistic children. We sought to clarify the direction of these associations longitudinally, and test executive function as a moderator. Participants were autistic children (N = 157; 15% female, 85% male) engaged in a longitudinal study. Analyses focused on two timepoints during preadolescence (M ages 9.7 and 10.7 years). A cross-lagged panel model tested whether parent-reported anxiety at age 9 years predicted teacher-reported social communication and interaction differences at age 10 years, and vice versa. Next, multigroup analyses tested for similarity in cross-lagged pathways at different levels of teacher-reported metacognition and behavioural regulation (two facets of executive function). At each time point, respectively, 22% and 21% of children had anxiety exceeding a suggested clinical threshold. Longitudinal associations between parent-reported anxiety and teacher-reported social communication and interaction differences were not significant in the full sample. However, multigroup analyses found lower levels of parent-reported anxiety at age 9 years predicted greater teacher-reported social communication and interaction differences at age 10 years among participants with clinically elevated behavioural dysregulation. Findings indicate low levels of anxiety, coupled with behavioural dysregulation, may signify potential for increasing social communication and interaction differences observed by teachers among autistic children entering adolescence.