Background: The optimal dose and fractionation scheme for stereotactic body radiotherapy (SBRT) is unknown. The biologic effects of ultra-high doses per fraction (>8Gy) are theoretical, but may include eliciting an effect on the endothelial cells of the tumor vasculature which could improve treatment response. This study aimed to determine the safety and maximally tolerated dose of 3-fraction SBRT for locally advanced pancreatic cancer (LAPC). Methods: A multi-site phase 1 dose escalation trial was conducted from March 2016 to April 2019 at Memorial Sloan Kettering Cancer Center (NCT02643498) and University of Colorado (NCT02873598). Patients with localized histologically confirmed pancreatic adenocarcinoma deemed unresectable on multidisciplinary review without distant progression following induction chemotherapy for ≥ 2 months were eligible. Patients received 3-fraction LINAC-based SBRT at 3 dose levels, 27Gy, 30Gy and 33Gy following a modified 3+3 design, allowing for enrollment of additional patients at the last dose level during the 90-day observation period, provided no dose-limiting toxicities (DLTs) were observed. DLTs were defined as ≥ Grade 3 treatment-related GI toxicity within 90 days of RT by CTCAE v.4. The secondary endpoints were overall survival (OS), local progression-free and distant metastasis-free survival (LPFS and DMFS, respectively). Univariate analysis using log-rank test was performed to identify factors associated with OS. Results: Twenty-three evaluable patients were enrolled, including 8 patients at 27Gy, 8 patients at 30Gy and 7 patients at 33Gy. The median age was 67 years (range 52 - 79), 9 patients (39%) were male, all were stage IIIwith a median tumor size of 3.5cm (range, 1.0 - 6.4) and CA19-9 of 60U/mL (range, <1 - 4880). All received chemotherapy for a median of 4.0 months (range 2.5 -11.4). There were no grade ≥ 3 abdominal pain, dyspepsia, diarrhea, nausea, vomiting, or gastrointestinal hemorrhage. Four patients underwent resections (pancreaticoduodenectomy=3, Appleby=1). Twelve-month rates of OS, DMFS and LPFS were 45.8 %, 37.7% and 53.0%, respectively. On univariate analysis, CA19-9 (HR=0.2365, 95%CI 0.07999 to 0.6990), but not dose level, size, N stage, tumor location, duration of chemotherapy were associated with OS. Twelve-month OS for patients with CA19-9 ≤ 60U/mL vs > 60U/mL were 80% vs 27% (p=0.0023). Conclusions: For select LAPC patients, dose escalation to the target dose of 33Gy in 3 fractions resulted in no DLTs and disease outcomes comparable to conventional RT. Lower pre-SBRT CA19-9 values were associated with improved OS and could help identify patients most likely to benefit from local therapies. Continued exploration of (ultra)hypofractionated schemes to maximize tumor control while enabling efficient integration of RT with systemic therapy is warranted. Clinical trial information: NCT02643498/NCT02873598.