Background: Patients with dMMR/MSI-H gastrointestinal (GI) cancer occasionally receive neoadjuvant immunotherapy (NIT) due to various clinical considerations. To systematically evaluate the outcome of these patients, we initiated the DREAM-GI national database, intending to harness real-world data to provide crucial insights into the outcomes, safety profile, and response patterns of dMMR/MSI-H GI cancer patients undergoing NIT. Herein, we present the initial findings. Methods: We developed a centralized database to collect de-identified clinical data from patients with dMMR/MSI-H GI cancers receiving NIT. We collected data retrospectively and prospectively through September 15, 2023. Results: The current report includes 50 patients with a median age of 67 years (range 32 to 90); 21 (42%) were female, 9 (18%) were Black, and the remaining were White. The cohort included the following tumor types- 31 (62%) colorectal, 7 (14%) gastroesophageal, and 12 (24 %) pancreaticobiliary. Most patients had localized disease (34, 68%), while the rest had oligometastatic disease. NIT consisted of pembrolizumab monotherapy in 40 (80%) patients, ipilimumab plus nivolumab in 8 (16%), and nivolumab plus chemotherapy in 2 (4%) patients. Most patients, 33 (66%), were treatment-naive. The median duration of NIT was 6 months (range: 1.5 to 55), with a time to best response of 3 months (range: 1.5 to 12). Among 47 evaluable patients, the best responses were as follows: an overall response rate of 75% (35/47), consisting of radiologic complete response (CR) in 20 (43%), pathologic CR in 4 (9%), and partial response in 11 (24%). The median progression-free survival and overall survival were not reached after a median follow-up of 14 months (range: 2-80), calculated from the date of the first immunotherapy dose to the last follow-up or death. Among 24 patients achieving CR, all remained progression-free after a median follow-up of 25.5 months (range: 3-80). Stable and progressive diseases were observed in 7 (15%) and 5 (10%) patients, respectively. Only 5 (10%) patients underwent surgery following NIT, with 4 (80%) achieving pathologic CR. The reasons for not proceeding with surgery were comorbidities in 40 (80%) patients, disease extent in 4 (8%), and patient refusal in 1 (2%). Out of the 12 patients who expired at the data cut-off, 1 was due to immunotherapy-related pneumonitis, 6 due to disease progression, and the remainder from underlying comorbidities. Conclusions: Our study underscores the remarkable response rates and durability of responses achieved with NIT in patients with dMMR/MSI-H GI cancers. Progression on NIT was infrequent. These real-world data support further investigation into non-operative approaches for patients with dMMR/MSI-H GI cancers.