AIMS: People with recently-diagnosed type 1 diabetes mellitus (T1D) may undergo a transient period of glycaemic control with less exogenous insulin. Identification of predictors of this 'remission' could inform a better understanding of glycaemic control. METHODS: Participants in the ADDRESS-2 study were included who had 1 or 2 assessments of remission status (coincident insulin dose and HbA1c measurement, with remission defined by ≤0.4 units insulin/kg-body-weight/day with HbA1c < 53 mmol/mol). Demographic and clinical presentation characteristics were compared according to remission status and predictors of remission were explored by logistic regression analysis. RESULTS: 1470 first and 469 second assessments of remission status were recorded within 12 months of diagnosis of T1D. Step increases in the probability of remission were identified at age-at-diagnosis 20 years and 3 months after diagnosis (both p < 0.001). Among those aged < 20 years, remission was associated with male gender (p = 0.02), no ketoacidosis (p = 0.02) and fewer than 2 symptoms at presentation (p = 0.004). None of these characteristics predicted remission in those aged ≥ 20 years. In the subgroup with two assessments, transition to remission was independently associated with first remission assessment in months 1-2 post-diagnosis (p = 0.01), with age-at-diagnosis ≥ 20 years (p = 0.01) and, in those aged < 20 years, with an early HbA1c of <57 mmol/mol. Adiposity, ethnicity, autoantibody status and other autoimmune disease were unrelated to remission. CONCLUSIONS: For those diagnosed before 20 years of age, males, ketoacidosis-free, with fewer symptoms and low early HbA1c were more likely to experience remission, but remission was most likely in anyone aged ≥ 20 at diagnosis.