Background: There are no accurate data on the diagnostic value of preoperative flexible bronchoscopy (FB) for persistent ground glass opacity (GGO) nodules. Methods: We retrospectively searched a database for subjects who had ‘GGO,’ ‘non-solid nodule,’ ‘part-solid nodule,’ or ‘sub-solid nodule’ in radiological reports of chest computed tomography between February 2004 and March 2012. Patients who had infiltrative GGO, focal GGO larger than 3 cm, or were lost to follow-up were excluded. We assessed the diagnostic value of preoperative FB in patients with persistent GGO nodules who underwent planned surgical resection. Results: In total, 296 GGO nodules were evaluated by FB in 264 patients with persistent GGO nodules who underwent preoperative FB and planned surgical resection. The median size of the GGO nodules was 18 mm, and 135 (46%) were pure GGO and 161 (54%) were mixed GGO. No visible tumor or unsuspected endobronchial metastasis was identified by preoperative FB. Only 3 (1%, 3/208) GGO nodules were identified preoperatively as malignant by bronchial washing cytology, which were all mixed GGO; no other accompanying etiology was identified by FB. Of all GGO nodules, 271 (91%) were subsequently confirmed as malignant and 25 (9%) were confirmed as benign by surgical resection. Consequently, the overall diagnostic sensitivity and negative predictive value of preoperative FB on a per-nodule basis were 1% (3/271) and 8% (25/293), respectively. The preoperative FB did not change the planned surgical strategy. Conclusions: Preoperative FB should not be routinely recommended in surgical candidates with persistent GGO nodules, especially pure GGO nodules.