There is limited research regarding the Gilliam Autism Rating Scales-3rd Edition (GARS-3) despite its extensive use. A comprehensive diagnostic evaluation, including the Autism Diagnostic Observation Schedule-2nd Edition (ADOS-2) was provided to 186 clinically referred children suspected of autism (\(\overline{X }\) age = 8.98; Autism [AUT] n = 87; Not Autism [NOT] n = 99). Mean difference analyses, Logistic Regressions, and ROC analyses were non-significant for both Autism Index scores. The author-suggested cutoff score of 70 correctly classified approximately 47% of participants, with false positive rates = 82.83–87.88%. ADOS-2 correlations were significantly lower vis-à-vis the standardization sample. The Social Interaction subscale demonstrated weak, marginal results, and sensitivity/specificity could not be optimized. In its current form, the GARS-3 does not demonstrate adequate criterion validity for use in assessment of complex community samples.