Abstract BACKGROUND Central nervous system tumours are the second most common childhood tumours and the largest cause of childhood cancer death in Australia. Tumours arise from many cell types within the developing and mature CNS, leading to a wide spectrum of clinical behaviour despite overlapping histological appearances. Emerging molecular testing methods allow more precise determination of cell lineage as well as genetic and epigenetic changes driving tumour development. Particularly since 2016, molecular findings have been incorporated into international classification schema, potentially impacting the frequency of individual diagnoses and their observed prognosis. METHODS We performed a retrospective cohort study including patients with newly diagnosed CNS tumours at Royal Children’s Hospital, Melbourne between 2011 and 2020 (total 473). The first and last 100 eligible patients in the data set were reviewed for diagnosis, survival and recorded use of molecular testing. Reviewed patients were assigned to one of two cohorts, Cohort 1 2011-2013 (pre-2016 WHO revision) or Cohort 2 2018-2020 (post-2016 WHO revision). RESULTS The frequency of diagnoses on biopsy rather than imaging alone was stable (81% Cohort 1 vs 83% Cohort 2). Molecular testing utilization increased markedly over time (10 molecular investigations in Cohort 1 vs 348 in Cohort 2). The incidence of astrocytic and oligodendroglial diagnoses was stable. The incidence of embryonal and neuronal/ mixed glial neuronal diagnoses increased in Cohort 2, and the incidence of ependymal diagnoses decreased (p=0.0122). There were no event free or overall survival differences between the whole of Cohorts 1 and 2, with numbers of individual diagnoses too small for meaningful comparisons over time. CONCLUSIONS Access to molecular testing has impacted on the pattern of reported diagnoses of central nervous system tumours in children. Although outcomes have not changed, it is hoped that over time, improved molecular understanding will lead to more effective therapy selection.