Purpose <p>Children born prematurely and those with medical and neurological complications are at increased likelihood of a diagnosis of autism. Accordingly, accurate screening and evaluation for autism in these populations are important to facilitate early intervention. Few autism diagnostic tools have been specifically validated for use in medically complex young children.</p> Methods <p>Eighty-eight children (Mean age = 25.82 months, <i>SD</i> = 5.23; 35 females) were administered an autism-specific caregiver interview (Toddler Autism Symptom Inventory [TASI]), semi-structured play and interaction observational assessment (TAP), and clinician rating scale (Childhood Autism Rating Scale, Second Edition [CARS2-ST]) as part of a targeted diagnostic consultation.</p> Results <p>The TASI, TAP, and CARS2-ST were highly sensitive (90–100%). Measures varied in specificity with the CARS2-ST demonstrating the strongest (100%), followed by the TAP (50–100%) and TASI (40–75%). Age, sex, prematurity, medical condition, and presence of a co-occurring neurodevelopmental condition impacted the validity of the TAP and TASI. The TASI and TAP symptoms/items accurately differentiated autistic and non-autistic children.</p> Conclusion <p>This study provides initial support for the validity of the TASI, TAP, and CARS2-ST in medically complex young children and supports a clinical best estimate approach to autism diagnosis, including history on the TASI and observations from the TAP to inform ratings on the CARS2-ST, by an experienced clinician in this patient population. While each tool demonstrated high sensitivity, specificity varied, highlighting the value of several sources of diagnostic information to improve diagnostic accuracy and facilitate timely identification of autism in toddlers with complex medical conditions.</p>