Objective: To evaluate item-factor structures of the 15 sensory items from the Survey of Pathways to Diagnosis and Services (Pathways) and examine the best fitting model. Methods: The study subjects were 1,968 children aged 6-17 years, who had ever been diagnosed with Autism Spectrum Disorder (ASD) and/or Intellectual Disability (ID). Factor analyses and item response theory models were used to determine the best fitting item-factor structure of the sensory items. The Strengths and Difficulties Questionnaires (SDQ) was used to test the concurrent validity of the sensory severity estimates. Results: A bifactor MIRT model (a general and four sensory factors) was selected as the best fitting model. All items statistically fitted to the bifactor model (p > .01) and showed moderate correlations with all five subscales of the SDQ (r s = .31 ~.51, p < .0001). The general sensory score differentiated the four different diagnostic groups (ASD, ID, ASD with ID, and no current symptoms) [F (3,1961) = 207.4, p < .0001]. Conclusion: The Pathways sensory items yielded reliable estimates of the general and each corresponding sensory factor by applying a bifactor MIRT model. The sensory score can be used as a valid sensory measure in the population survey.