While the overall cure rate for paediatric acute lymphoblastic leukaemia (ALL) is now approaching 90%, patients with high-risk subtypes or those who relapse remain difficult to treat. Recent “omics” approaches have identified novel paediatric ALL subtypes and highlighted the heterogeneity within these subtypes. This is exemplified by the distinct heterogeneity within recently identified high-risk ALL subtypes such as Ph-like (BCR-ABL1-like) ALL and early T-cell precursor (ETP) ALL, all of which are likely to require different approaches to treatment. This heterogeneity should be considered in the development and preclinical testing of novel therapies for high-risk or relapsed ALL. The highly immune-deficient mouse strains that have become available over the past 15-20 years (e.g. NOD/SCID, NSG) have been shown to allow the development of patient-derived xenografts (PDXs, or avatars) from all ALL subtypes and almost every ALL patient. Our research program has developed panels of molecularly annotated PDXs from >100 patients, with all PDXs derived directly from bone marrow or peripheral blood biopsies without prior in vitro culture. The PDX sub-panels are representative of the heterogeneity of paediatric ALL in general, as well as specific high-risk subtypes. This presentation will focus on how PDX models can be utilised to improve the treatment of childhood leukaemia in specific leukaemia subgroups, as well as in personalised approaches to treatment of patients in real-time.