The present study evaluates the test-retest reliability of six theory of mind (ToM) tasks that measured cognitive, affective, and spontaneous ToM in 7 to 11 year-old children with autism spectrum disorder. Our results revealed considerable variation in test-retest reliability depending on the type of ToM task, which ranged from poor to good with the majority of the measures exhibiting moderate reliability. Results inform which common measures of cognitive ToM should be selected versus avoided in future intervention work, suggest our measure of spontaneous ToM should be used more widely in intervention and ToM research more broadly, and indicate more work is needed to develop reliable measures of affective ToM. Implications for research and clinical practice are discussed.