Background: Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is generally performed for the diagnosis of hilar/mediastinal lymph node metastasis in patients with lung cancer. Recently, a 25-gauge (G) needle became available, but robust evidence of its usefulness in routine clinical practice is still lacking.Methods: A prospective randomized crossover trial was performed, in which patients with suspected hilar/ mediastinal lymph node metastasis of lung cancer underwent EBUS-TBNA. The primary endpoint was the rate of yield histology specimens containing malignant cells.Results: From December 2018 to February 2020, 102 patients were randomly assigned to EBUS-TBNA using a 22G needle first, followed by a 25G needle (n=50) or EBUS-TBNA using a 25G needle first, followed by a 22G needle (n=52). There was no difference in the diagnostic yield of malignancy between the histology specimens obtained by using the 22G and 25G needles (75% vs. 75%, respectively, P=0.37). The sizes of the tissue samples (16.4 vs. 4.9 mm 2 , respectively) and number of malignant cells in the tissue samples (626 vs. 400, respectively) were both significantly higher when using the 22G needle than when using the 25G needle.Conclusions: No significant difference in the diagnostic yield between the 22G and 25G needles was observed for the diagnosis of lymph node metastasis of lung cancer, suggesting that needles of either gauge could be used for the biopsy. However, we would recommend use of the 22G needle, because it provided larger specimens and specimens containing larger numbers of malignant cells.