Background: Immune checkpoint inhibitor (ICI)-mediated colitis (IMC) is a common and serious adverse event. Although small series have described the clinical presentation of IMC, large multicenter series that integrate clinical, endoscopic, and histologic findings are lacking. Methods: We retrospectively assessed patients who received ICI and had endoscopically confirmed IMC from 2010 to 2019. IMC was graded based on the CTCAE version 5.0 criteria. Multivariate logistic regression analyses were conducted to assess factors associated with recurrence of IMC symptoms and long duration of corticosteroids use (> 70 days). Results: 675 patients were included. 387 patients were males (57%). Median age was 63 years. Melanoma was the most common cancer type (327; 48%). Most (365; 54%) patients received CTLA-4 inhibitor ICI, as monotherapy or in combination with PD-(L)1. Median time from ICI therapy to IMC was 62 days. IMC was grade 2 in 335 (50%) patients, Grade 3 in 181 (27%), and grade 4 in 16 (3%). 155 (23%) patients had mucosal ulceration on endoscopy, 91 of them had severe features (deep, large, or multiple ulcers); 336 (50%) patients had non-ulcerative inflammation. The rest had normal endoscopic findings with histologic inflammation. Most patients were admitted to the hospital for management of IMC (405; 60%) and 16 (3%) needed ICU-level of care. Treatment included corticosteroids in 577 (85%) patients (median duration 52 days), TNF inhibitor in 245 (36%), and vedolizumab in 90 (13%). 202 (32%) patients had recurrent IMC after resolution of symptoms. On multivariate logistic regression, factors associated with IMC recurrence and long (> 70 days) duration of corticosteroid therapy were grade of IMC (p = 0.049), treatment with infliximab or vedolizumab (p = 0.044), presence of mucosal ulceration (p = 0.034), or features of active histologic inflammation (p = 0.076). Of note, patients with mucosal ulceration received infliximab or vedolizumab more frequently (p< 0.001). For patients with grade 2 IMC, mucosal inflammation on endoscopy and delay in performing endoscopy with time from IMC onset to endoscopy more than a month were associated with IMC recurrence and longer duration of corticosteroid use (p = 0.029 and p< 0.001, respectively). 16 (3%) patients had colonic perforation, 7 of them underwent surgical resection. No IMC-related death occurred. Conclusions: IMC is a clinically significant adverse event that can lead to premature termination of ICI therapy with high rates of hospital admission. Rarely, it results in colonic perforation requiring surgical intervention and ICU admission. Our data suggest that there is a utility of endoscopic and histologic evaluation in the prediction of worse outcomes from IMC. This finding is particularly important for grade 2 IMC as current guidelines do not recommend endoscopic evaluation for this group.