The primary aim of this article was to provide a structural validation of the 28-item Autism Spectrum Quotient–Short Form questionnaire in a sample of adults with clinically diagnosed autism spectrum disorders (n = 148). Confirmatory factor analysis demonstrated that the proposed structure, comprising a second-order Social Skills factor subsuming first-order factors of Social Skills, Routine, Switching and Imagination, and a Numbers/Patterns factor showed good acceptable fit to the data (² = 464.27(345), p < .05; comparative fit index = .91; Tucker Lewis index = .90; root mean squared error of approximation = .048; weighted root mean square residual = 0.98). Summed mean scores for the total sample, and males and females separately, showed no significant differences to previously reported means, with the exception of Numbers/Patterns, for which the current sample reported significantly lower means in both males (t = 3.37, p < 0.001) and females (t = 2.33, p < 0.05). The results provide support for both the proposed structure of the AQ-S and the stability of total and subscale mean scores in different samples of participants with autism spectrum disorder diagnoses.