Background: Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized by impairment in social interactions and communication. Additional features include restricted, repetitive patterns of behaviors, and differences in sensory processing. The clinical presentation of patients with ASD is heterogeneous, likely reflecting multiple underlying etiologies. Heterogeneity in presentation and treatment response are barriers to development of precise therapeutic approaches. Therefore, identification of clinically meaningful subgroups within ASD is critical to develop targeted interventions. We hypothesized that sensory features can be used to identify clinically recognizable subgroups with shared underlying etiologies. Methods: Subjects included 378 individuals with a clinical diagnosis of ASD who contributed Short Sensory Profile (SSP) data assessing the frequency of sensory behaviors and whole genome sequencing results to the Autism Speaks’ MSSNG database. To determine if the SSP could be used to subgroup individuals with ASD, we performed cluster analysis on responses to all 38 questions, followed by an independent cluster analysis using only a subset of questions selected specifically to assay hyper- and hypo-sensitivity to sensory stimulation. Cross-validation of the resulting clusters determined the final subgroups. To test for shared underlying etiologies, we correlated variant frequency across subgroups for each of 24,896 genes. Variant frequency included any variation in each gene regardless of the type of variant. To be significantly associated with a subgroup, a gene variant frequency had to be greater than four standard deviations (SD) from the mean frequency for all subgroups and 3 SD different from each subgroup.Results: We identified seven distinct sensory-based ASD subgroups. Subgroup 1, characterized by atypical scores in all sensory areas, was not associated with any genes. Subgroups 2, 4 and 6 were significantly associated with four to six genes each. Subgroups 3, 5 and 7 were enriched for 126, 12 and 50 genes, respectively. Limitations: This study was performed using retrospective data that did not include other phenotypic data such as age, comorbidities, or measures of disease severity. All those likely contribute to the variability of the identified subgroupsConclusions: These results support the use of sensory features to identify ASD subgroups with shared genetic mechanisms.