In patients with pancreatic ductal adenocarcinoma (PDAC), optimal treatment selection, including multimodality regimens such as neoadjuvant chemoradiotherapy (NACRT), can be clinically transformative. Unfortunately, currently no predictive biomarkers are available that can guide the use of NACRT in PDAC patients. Accordingly, herein we developed a novel gene signature that can preoperatively predict NACRT‐sensitivity in PDAC patients. Herein, we evaluated the performance of a 10‐gene panel in 749 PDAC cases, which included two public datasets (The Cancer Genome Atlas and International Cancer Genome Consortium; n = 276), and three clinical specimen cohorts (n = 417), and a pre‐NACRT endoscopic ultrasound‐guided fine‐needle aspiration (EUS‐FNA) biopsy cohort (n = 56). The potential predictive performance of this signature was evaluated and compared to CA‐19‐9 levels and key clinicopathological factors. We first evaluated the prognostic potential of a 10‐gene panel which significantly predicted overall survival in both public datasets (P < .01, P < .01), and two in‐house patient cohorts (P < .01, P = .04). In the pre‐NACRT EUS‐FNA cohort, we established a radio‐sensitivity gene panel (RSGP) which yielded highly robust (area under the curve AUC = 0.91; 95% CI: 0.81‐0.97) for predicting response to gemcitabine‐based NACRT. Multivariate logistic regression analysis revealed that RSGP was an independent predictor for response to NACRT (OR = 2.70; 95% CI: 1.25‐5.85), and this response‐prediction was even more robust when CA‐19‐9 levels were included into the model. In conclusion, we have validated and developed a novel gene signature that is highly robust in predicting response to NACRT, even in preoperative settings, highlighting its clinical significance for optimizing and personalizing treatment strategies in PDAC patients. What's new? While the benefits of radiation treatment have been well established in solid cancers, the criteria for optimal indication of radiotherapy in pancreatic ductal adenocarcinoma and the best regimens for its clinical application remain unclear. The availability of biomarkers that could help predict response to radio‐sensitivity before treatment could lead to better‐informed decision‐making. Here, the authors report developing and validating a novel gene expression signature that is highly robust in predicting patient response to neoadjuvant chemoradiotherapy, even in preoperative settings. The results highlight the clinical significance of this gene signature for optimizing and personalizing treatment strategies in patients with pancreatic cancer.