Previous research has demonstrated mixed evidence on impaired memory functioning in autism spectrum disorder (ASD), with the only consensus that there appears to be much heterogeneity. In addition, no research to date has examined the stability of memory in ASD. This study examined the stability of memory function in ASD compared to typically developing age-matched controls. Participants were administered the Test of Memory and Learning (TOMAL) twice, three years apart, in an established longitudinal NIH-supported investigation of ASD. Based on available research contrasting memory development in healthy individuals versus those with ASD, it was hypothesized that memory performance in the control group would be stable across time and that, compared to the control group, the autism group would demonstrate less stable memory function as measured by the TOMAL. Repeated Measures ANOVA and Reliable Change Index calculations of TOMAL Index and Subtest scores largely supported these hypotheses. The control group remained stable across time on all TOMAL indices and the ASD group showed improvement on the Composite Memory Index, Verbal Memory Index, and Delayed Memory Index but not the Non Verbal Memory Index. Clinical and research implications are discussed.