Gao et al. established 1,075 patient-derived xenograft (PDX) models from a range of human solid tumour samples and carried out a PDX clinical trial with 62 treatments. Their analysis correlated genomic alterations with responses to therapy, confirming relationships that have already been established and thus validating the approach. The authors also found some novel combinations that could be explored in more focused trials. Therefore, PDX models may offer more translational reproducibility than traditional preclincial approaches that use cell lines.