Anxiety‐related difficulties are common in ASD, but measuring anxiety reliably and validly is challenging. Despite an increasing number of studies, there is no clear agreement on which existing anxiety measure is more psychometrically sound and what is the factor structure of anxiety in ASD. The present study examined the internal consistency, convergent, divergent, and discriminant validity, as well as the factor structure of the Spence Children's Anxiety Scale‐Parent Version (SCAS‐P), in a large international pooled sample of 870 caregivers of youth with ASD from 12 studies in the United Kingdom, United States, and Singapore who completed the SCAS‐P. Most were community recruited, while the majority had at least one measure of ASD symptomatology and either cognitive or adaptive functioning measures completed. Existing SCAS‐P total scale and subscales had excellent internal consistency and good convergent, divergent and discriminant validity similar to or better than SCAS‐P properties reported in typically developing children, except for the poorer internal consistency of the physical injury subscale. Confirmatory Factor Analysis (CFA) of the existing SCAS‐P six‐correlated factor structure was a poor fit for this pooled database. Principal component analysis using half of the pooled sample identified a 30‐item five correlated factor structure, but a CFA of this PCA‐derived structure in the second half of this pooled sample revealed a poor fit, although the PCA‐derived SCAS‐P scale and subscales had stronger validity and better internal consistency than the original SCAS‐P. The study's limitations, the use of the SCAS‐P to screen for DSM‐derived anxiety problems in ASD and future research directions are discussed. Autism Res 2017, 10: 1629–1652. © 2017 International Society for Autism Research, Wiley Periodicals, Inc.