The Strengths and Difficulties Questionnaire (SDQ) is used to measure psychopathological symptoms in children and adolescents from 4 to 17 years old, but its underlying structure is still a matter of debate. Indeed, on the basis of a systematic review of English and non-English articles conducted using multiple databases, 54 studies reporting on the factor structure of the SDQ were located. The original 5 first-order factor structure is generally supported by exploratory procedures, but support based on confirmatory factor analyses is not clear. We analysed data from 889 youths from the general French population, rated on the SDQ by their teachers. We tested the original model, hierarchical models and bifactor models. The best-fitting model is a bifactor model with the five a priori factors grouped in two global factors (Externalizing Disorders-Hyperactivity and Conduct-and Internalizing Disorders-Peer relationships and Emotions) and one Strength/Prosocial factor. However, we show that the Conduct-Specific factor should not be used in practice in its current state, that the Hyperactivity-Specific factor mainly covers hyperactivity rather than inattention, and that the Peer Problems-Specific factor mainly reflects a preference for solitude. Nevertheless, the measurement model proved to be fully invariant across gender and school levels (kindergarten, primary and secondary schools), with statistically significant differences in latent means between genders only. Beyond computing the five a priori scores when using the teacher ratings of the SDQ, our results prove the usefulness of computing Externalizing Disorders and Internalizing Disorders global scores.