Background: The prevalence of immune-mediated toxicities from immune checkpoint inhibitors (ICIs) is well described. However, the characteristics and treatment patterns for patients with emergency presentations due to immune-mediated toxicity are less well known. Methods: This study of all emergency presentations in patients treated with ICIs was performed at The Christie NHS Foundation Trust, Manchester, United Kingdom from May 2018-February 2020. The aims were to describe the patient and treatment characteristics of those diagnosed with an immune-mediated toxicity. Results: In total, 597 patients receiving ICIs had an emergency presentation and 191/597 (32%) were diagnosed with an immune-mediated toxicity. Of these patients, the median age was 64 years and 127/191 (67%) were male. The most common tumour types were melanoma (53%) and lung (22%) and the most common ICI received was ipilimumab + nivolumab combination immunotherapy (42%), followed by pembrolizumab monotherapy (21%) and nivolumab monotherapy (20%). The median number of cycles received was 3 (range 1-54), and 73/191 (38%) previously had ≥ grade 2 immune-mediated toxicity. The most common diagnoses were colitis (38%), hepatitis (15%), and pneumonitis (14%). The majority, 180/191 (94%) received steroids and 52/180 (29%) patients required second-line immunosuppression. The most common second-line immunosuppressants used were mycophenolate mofetil (58%) and infliximab (50%). Eleven patients (22%) required more than one second-line immunosuppressant. Conclusions: The majority of patients with emergency presentations due to immune-mediated toxicity were being treated with combination immunotherapy for melanoma. More than a third of patients had previous ≥ grade 2 immune-mediated toxicity. Over one quarter of patients treated with steroids required second-line immunosuppression. Identifying these characteristics can help inform which patients receiving ICIs seeking medical review need admission to a center with experience in managing immune-mediated toxicity.