Study objectives Endobronchial ultrasonography (EBUS) is used as guidance for transbronchial needle aspiration (TBNA), and real-time imaging of the needle position cannot be confirmed with a single-channel bronchoscope. We assessed the usefulness of EBUS-guided TBNA using a double-channel bronchoscope (EBUS-D), which provides real-time needle position, and compared it with EBUS-guided TBNA using a single-channel bronchoscope (EBUS-S). Design Randomized, comparative prospective study. Setting Hiroshima City Hospital, a tertiary-referral teaching hospital. Patients Between January 2000 and August 2003, 55 patients with intrathoracic lymphadenopathy were included. Patients were randomized to undergo EBUS-D (n = 30) or EBUS-S (n = 25). Methods EBUS-D: The EBUS probe and TBNA catheter were inserted simultaneously through a double-channel bronchoscope. Once the needle placement in the lesion was confirmed by EBUS, TBNA was performed. EBUS-S: The EBUS probe was removed after the determination of the penetration site. Then, the TBNA catheter was inserted and TBNA was performed. Results All the lymph nodes could be visualized with EBUS in each group of patients. In the EBUS-D group, the TBNA needle was visualized as a hyperechoic point on the real-time EBUS image. The diagnostic accuracy rate of EBUS-D and EBUS-S were statistically significantly different (97% vs 76%, respectively; p=0.025). On second attempt of TBNA, the diagnostic rate of the EBUS-D group was superior to that of the EBUS-S group (85.7% vs 33.3%, respectively; p=0.036). The mean number of penetrations was 1.24 in the EBUS-D group and 1.36 in the EBUS-S group. No complications were observed in the EBUS-D group, but a self-limiting hemorrhage occurred in a patient in the EBUS-S group. Conclusion EBUS-D is useful for diagnosing intrathoracic lymphadenopathy, and the obtained specimen with real-time confirmation of the needle is directly proportional to an accurate diagnosis.