The 2016 WHO classification of tumours of the central nervous system represents the new paradigm among the specialists in the brain tumours and proposes a new approach combining histopathological and molecular features into diagnosis named ‘integrated diagnosis’. The aim of this challenge is to overstep the interobserver variability of diagnosis based on previous classifications in order to ensure homogenous biological entities with a more accurate clinical significance. Over the last two decades, several molecular aberrations into gliomagenesis were highlighted and then confirmed as emerging biomarkers through prognostic stratification. In particular, IDH1/IDH2 genes mutations, 1p/19q codeletion and mutations in genes encoding histone H3 variants drastically changed the knowledge about diffuse gliomas inducing the WHO working group to consider the phenotype-genotype approach. In the present review, the historical development of the diagnosis of brain tumours from the 3D spatial configuration to the integration of multidisciplinary data up to recent molecular alterations is discussed. At the national level, the RENOCLIP network (supported by the National Cancer Institute) contributes to improve the standardization of histological diagnosis and the facilitation of access to molecular biology platforms for the detection of genetic aberrations necessary for integrated diagnosis. Importantly, the French POLA cohort allowed to test the clinical impact of the new criteria introduced by 2016 WHO classification of CNS tumours confirming the high accuracy in predicting clinical behaviour for diffuse gliomas.Le texte complet de cet article est disponible en PDF.