Background/purpose: Research biopsies (RBs) are crucial for developing novel molecular targeted agents. However, the safety and diagnostic yields of RBs have not been investigated in EGFR mutation-positive non-small cell lung cancer (NSCLC) patients pretreated with epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs). Methods: We searched the medical records of NSCLC patients who participated in lung cancer clinical trials and underwent mandatory RBs between 2012 and 2014 at our institution. Only patients with EGFR mutation-positive NSCLC pretreated with at least 1 EGFR-TKI were enrolled. Results: Of 140 enrolled patients, 73 (52.1%) and 59 (42.1%) had exon 19 deletions and exon 21 L858R mutation, respectively. Before RBs, 108 (77.1%), 83 (59.3%), and 36 (25.7%) patients had been treated with gefitinib, erlotinib, and afatinib, respectively. Computed tomography-guided percutaneous core needle biopsy was the most frequently used modality among 181 RBs performed (50.8%), followed by ultrasonography-guided (32.0%) and endoscopic RBs (16.0%). The most common RB sites were the lung (69.6%), pleura (8.8%), and liver (6.1%). Pathologic examinations revealed malignant cells in most RB specimens (72.9%). Complications due to RBs included pneumothorax (11.6%), bleeding (6.1%), and infection (1.1%). Only 1 patient required chest tube placement for pneumothorax, and 2 patients underwent endotracheal intubation because of bleeding. Conclusion: RBs in this patient population were generally safe. Pneumothorax was the most frequent complication; bleeding, while infrequent, increased the risk of severe events. The diagnostic yields and complications of any particular modality should therefore be discussed with prospective clinical trial participants. Keywords: Complications, Computed tomography-guided percutaneous core needle biopsy, EGFR mutation, Non-small cell lung cancer, Research biopsies