Background: This systematic review aimed to determine the accuracy of imaging modalities to predict resectability and R0 resection for borderline resectable (BRPC) or locally advanced pancreatic cancer (LAPC) after neoadjuvant therapy (NAT). Methods: A systematic search of major databases was performed in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Results: Fifteen studies identified 995 patients of which 683 had BRPC and 312 LAPC. Computed tomography (CT) scan was the most common modality for re-staging (n = 14), followed by positron emission tomography (PET)-CT (n = 3) and endosonography (EUS) (n = 2). Stable disease on RECIST criteria was found in 67% of patients (range 53–80%) with 20% demonstrating reduction in tumour size. A total of 60% of patients underwent surgery post-NAT (range 31–85%) with a R0 rate of 88% (range 57–100%). Accuracy for predicting R0 resectability and T-stage on CT scan was 71 and 49%. A reduction in SUV max on PET-CT and reduction of tumour stiffness on EUS elastography positively correlated with resectability. Conclusions: More than half the patients undergo resection post-NAT for LAPC and BRPC. Stable, or reduction of, tumour disease may predict resectability. Reduction in tumour SUV max on PET-CT and decreased tumour stiffness on EUS elastography may be potential markers of NAT response and resectability.