We report on initial patient studies to evaluate the performance of a commercial respiratory gating radiotherapy system. The system uses a breathing monitor, consisting of a video camera and passive infrared reflective markers placed on the patient's thorax, to synchronize radiation from a linear accelerator with the patient's breathing cycle. Six patients receiving treatment for lung cancer participated in a study of system characteristics during treatment simulation with fluoroscopy. Breathing synchronized fluoroscopy was performed initially without instruction, followed by fluoroscopy with recorded verbal instruction (i.e., when to inhale and exhale) with the tempo matched to the patient's normal breathing period. Patients tended to inhale more consistently when given instruction, as assessed by an external marker movement. This resulted in smaller variation in expiration and inspiration marker positions relative to total excursion, thereby permitting more precise gating tolerances at those parts of the breathing cycle. Breathing instruction also reduced the fraction of session times having irregular breathing as measured by the system software, thereby potentially increasing the accelerator duty factor and decreasing treatment times. Fluoroscopy studies showed external monitor movement to correlate well with that of the diaphragm in four patients, whereas time delays of up to 0.7 s in diaphragm movement were observed in two patients with impaired lung function. From fluoroscopic observations, average patient diaphragm excursion was reduced from 1.4 cm (range 0.7-2.1 cm) without gating and without breathing instruction, to 0.3 cm (range 0.2-0.5 cm) with instruction and with gating tolerances set for treatment at expiration for 25% of the breathing cycle. Patients expressed no difficulty with following instruction for the duration of a session. We conclude that the external monitor accurately predicts internal respiratory motion in most cases; however, it may be important to check with fluoroscopy for possible time delays in patients with impaired lung function. Furthermore, we observe that verbal instruction can improve breathing regularity, thus improving the performance of gated treatments with this system.