Purpose <p>This study examined which DSM-5-TR criteria most strongly predict autism diagnosis in adults and whether predictive patterns differ by sex assigned at birth.</p> Methods <p>A team of neurodivergent researchers designed and carried out the study with input from autistic psychologists and community members. The participants were English-speaking adults who sought telehealth services from an autism mental health service provider. A convenience sample of 234 adults (mean age 34.25 years, 72.2% assigned female at birth) underwent an autism evaluation through a telehealth organization. Licensed psychologists rated seven DSM-5-TR domains on a 0–2 scale based on structured interviews and standardized measures. Ridge penalized logistic regression was used.</p> Results <p>Results indicated that social-emotional reciprocity (OR = 5.21) and nonverbal communication (OR = 4.82) were the strongest predictors of autism diagnosis, followed by relationship differences (OR = 3.63), need for routines (OR = 2.57), and repetitive behaviors (OR = 2.20). Intense interests and sensory processing differences showed limited predictive utility. Sex assigned at birth did not meaningfully enhance diagnostic prediction beyond core symptom presentation, except for a modest interaction effect with nonverbal communication in individuals assigned female at birth.</p> Conclusions <p>These findings suggest clinicians may benefit from prioritizing assessment of social-emotional reciprocity and nonverbal communication domains when evaluating adults for autism, potentially improving diagnostic efficiency and accuracy.</p>