OBJECTIVE Cerebrospinal fluid (CSF) rhinorrhea is among the most common complications following transsphenoidal surgery for sellar region lesions. The aim of this study was to review the authors' institutional experience in identifying, repairing, and treating CSF leaks associated with direct endonasal transsphenoidal operations. METHODS The authors performed a retrospective review of cases involving surgical treatment of pituitary adenomas and other sellar lesions at the University of Southern California between December 1995 and March 2016. Inclusion criteria included all pathology of the sellar region approached via a direct microscopic or endoscopic endonasal transsphenoidal approach. Demographics, pathology, intraoperative and postoperative CSF leak rates, and other complications were recorded and analyzed. A literature review of the incidence of CSF leaks associated with the direct endonasal transsphenoidal approach to pituitary lesions was conducted. RESULTS A total of 1002 patients met the inclusion criteria and their cases were subsequently analyzed. Preoperative diagnoses included pituitary adenomas in 855 cases (85.4%), Rathke's cleft cyst in 94 (9.4%), and other sellar lesions in 53 (5.2%). Lesions with a diameter ≥ 1 cm made up 49% of the series. Intraoperative repair of an identified CSF leak was performed in 375 cases (37.4%) using autologous fat, fascia, or both. An additional 92 patients (9.2%) underwent empirical sellar reconstruction without evidence of an intraoperative CSF leak. Postoperative CSF leaks developed in 26 patients (2.6%), including 13 (1.3% of the overall group) in whom no intraoperative leak was identified. Among the 26 patients who developed a postoperative CSF leak, 13 were noted to have intraoperative leak and underwent sellar repair while the remaining 13 did not have an intraoperative leak or sellar repair. No patients who underwent empirical sellar repair without an intraoperative leak developed a postoperative leak. Eight patients underwent additional surgery (0.8% reoperation rate) for CSF leak repair, and 18 were successfully treated with lumbar drainage or lumbar puncture alone. The incidence of postoperative CSF rhinorrhea in this series was compared with that in 11 other reported series that met inclusion criteria, with incidence rates ranging between 0.6% and 12.1%. CONCLUSIONS In this large series, half of the patients who developed postoperative CSF rhinorrhea had no evidence of intraoperative CSF leakage. Unidentified intraoperative CSF leaks and/or delayed development of CSF fistulas are equally important sources of postoperative CSF rhinorrhea as the lack of employing effective CSF leak repair methods. Empirical sellar reconstruction in the absence of an intraoperative CSF leak may be of benefit following resection of large tumors, especially if the arachnoid is thinned out and herniates into the sella.