Oesophageal cancer is the 8th most common cancer worldwide and has the 6th highest mortality rate of all cancers. The 5-year survival rate following oesophageal adenocarcinoma (OAC) diagnosis is dismal at <15%, indicating a dire need for improved therapeutic strategies and early detection. OAC develops stepwise following exposure to chronic gastric reflux: From pre-malignant Barrett’s metaplasia, through stages of low- and high-grade dysplasia until developing into invasive cancer. Mutation or loss of common tumour suppressor genes TP53 and SMAD4 act as markers for cancer progression, occurring in high-grade dysplastic tissue and invasive OAC, respectively. Our novel in vivo tumourigenesis model demonstrates progression of Barrett’s metaplasia to OAC, in which SMAD4-deficient Barrett’s metaplasia cells form tumours in immunodeficient mice after a period of latency and in a dose-dependent manner. This delayed tumour growth onset suggests further drivers are required for oncogenesis, and these SMAD4-deficient cells and tumours display a greater degree of genomic instability than wildtype-SMAD4 controls. A genome-wide CRISPR-Cas9 knockout screen unveiled a synthetic lethal relationship between SMAD4-deficiency and cell cycle checkpoint inhibition, suggesting a role for SMAD4 in maintaining genomic stability and a potential novel therapeutic avenue for SMAD4-deficient OAC. Additionally, a concurrent in vivo CRISPR-Cas9 tumourigenesis screen produced tumours 4-fold faster than the previous model and identified regulators of mTOR signalling as co-operative drivers of tumourigenesis in OAC. Wildtype-SMAD4 cells failed to generate tumours despite undergoing the same genetic perturbations, indicating a potential gatekeeping effect of SMAD4 in mTOR-mediated OAC tumourigenesis. In sum, loss of SMAD4 acts as a double-edged sword, increasing genomic instability and thereby rendering OAC cells sensitive to cell cycle checkpoint inhibition, whilst simultaneously co-operating with modulated mTOR signalling to promote tumourigenesis in OAC xenograft models.