Logistic regression was used to examine the use of Autism Spectrum diagnostic categories from pre-COVID-19 in-person evaluations and COVID-19 telehealth evaluations at a specialist community mental health clinic. The diagnostic classification for children 0–5 (DC: 0–5) affords a wider range of diagnoses that allowed for inferences of clinician certainty of diagnosis. Use of full criteria diagnoses was significantly lower from telehealth evaluations during the pandemic, and was less certain for younger children, some non-English speakers, and children reporting Native American/Alaska Native race. Higher Child Behavior Checklist (CBCL) ASD subscale scores, lower CBCL total scores, and global developmental delay diagnoses predicted greater use of full ASD diagnoses. Findings suggest factors that could identify children appropriate for telehealth evaluations.