Abstract: Background: Personalised medicine strategies may improve outcomes in pancreatic ductal adenocarcinoma (PDAC), but validation of predictive biomarkers is required. Having developed a clinical trial to assess the ATR inhibitor, AZD6738, in combination with gemcitabine (ATRi/gem), we investigated ATM loss as a predictive biomarker of response to ATRi/gem in PDAC. Methods: Through kinase inhibition, siRNA depletion and CRISPR knockout of ATM, we assessed how ATM targeting affected the sensitivity of PDAC cells to ATRi/gem. Using flow cytometry, immunofluorescence and immunoblotting, we investigated how ATRi/gem synergise in ATM-proficient and ATM-deficient cells, before assessing the impact of ATM loss on ATRi/gem sensitivity in vivo. Results: Complete loss of ATM function (through pharmacological inhibition or CRISPR knockout), but not siRNA depletion, sensitised to ATRi/gem. In ATM-deficient cells, ATRi/gem-induced replication catastrophe was augmented, while phospho-Chk2-T68 and phospho-KAP1-S824 persisted via DNA-PK activity. ATRi/gem caused growth delay in ATM-WT xenografts in NSG mice and induced regression in ATM-KO xenografts. Conclusions: ATM loss augments replication catastrophe-mediated cell death induced by ATRi/gem and may predict clinical responsiveness to this combination. ATM status should be carefully assessed in tumours from patients with PDAC, since distinction between ATM-low and ATM-null could be critical in maximising the success of clinical trials using ATM expression as a predictive biomarker.