BACKGROUND: Diabetes insipidus (DI) is a well-known complication of transsphenoidal surgery. However, the risk factors for DI remain controversial. METHODS: We conducted a retrospective study of patients who underwent endoscopic transsphenoidal surgery for pituitary adenoma at our institution during a 5-year period. The patients were divided into a DI group and a non-DI group. Logistic regression analyses were used to identify risk factors for postoperative DI. In subgroup analysis, the DI group was divided into transient DI and permanent DI groups, and perioperative factors were compared between groups. RESULTS: Of 101 patients, 58 were in the non-DI group (57.4%) and 43 were in the DI group (42.6%). Permanent DI occurred in 7 patients (6.9%). In univariate analyses, statistically significant risk factors were suprasellar extension, tumor functionality, and intraoperative cerebrospinal fluid leaks by Esposito grade. In multivariate logistic regression analysis, Esposito grade was the only statistically significant risk factor (P = 0.015). The frequency of DI increased as the Esposito grade increased (P = 0.0002 for the trend). In subgroup analysis, postoperative nadir sodium concentration was lower in the permanent DI group (128.1 ± 2.78 mmol/L) than in the transient DI group (135 ± 1.22 mmol/L; P = 0.035), and the optimal cutoff value was 124.5 mmol/L, with a sensitivity of 57.1% and a specificity of 91.7% (area under the curve = 0.76, P = 0.034). CONCLUSIONS: Intraoperative cerebrospinal fluid leak by Esposito grade is associated with postoperative DI. These data can be applied to help identify high-risk patients who need more aggressive follow-up and fluid management.