There is a need to investigate the diagnostic utility of autism diagnostic questionnaires in school-age children, who are increasingly being referred for autism assessment. Aside from the standardization sample, little research has been conducted on the Autism Spectrum Rating Scales, particularly regarding teacher reports. This study investigated the criterion validity of the Autism Spectrum Rating Scales-Teacher report for ages 6–18 (ASRS-T6-18) in a well-characterized community sample of 409 children (autism [AUT] n = 122; non-autism [NOT] n = 287; [Formula: see text] age = 9.91) evaluated with the Autism Diagnostic Observation Schedule-2, a well-validated autism diagnostic measure. Significant mean differences with small to moderate effect sizes ( d = 0.28–0.56) were observed for all scores except Self-Regulation, Adult interaction, and Attention. Logistic Regression and Receiver Operating Characteristic analyses conducted for the Summary, ASRS, and two Treatment scales (Peer and Social-Emotional Reciprocity) indicated low ORs (<|1.08|) and AUCs in the poor range (<.67). Total and Unusual Behavior emerged as comparatively stronger scales with covariates (age and IQ); other scales were generally similar with/without control variables. Sensitivity and specificity could not be optimized. The suggested 60 t-cutpoint had fair or good sensitivity (76.23–83.61) for all examined scales except Unusual Behavior (68.03). However, specificity was poor (46.93–54.36), with high false positive rates (45.64%–63.07%). In complex community samples, the ASRS-T6-18 behaves more like a screener as opposed to a diagnostic measure.