Temporal changes in respiration could influence navigator-echo (NE)-gated MR coronary angiography (MRCA), but systematic investigation of the effects of such variations and how to limit them has not been performed. We addressed these issues by studying the influence of time in the magnet on diaphragm position and respiratory patterns using NE diaphragm monitoring in volunteers and a phantom model. NE diaphragm monitoring was performed at .5 T in 10 subjects over a total period of 35 minutes. The end-expiratory position was sustained for longer (1.1 vs .4 seconds, P < .001) and with greater position stability (SD 1.9 vs 5.9 mm, P = .01) than the end-inspiratory position. Drift of the end-expiratory position occurred over time, causing a fall in scan efficiency (44-28%, P = .01). Up-drift of the end-expiratory position was most common. Loss of scan efficiency was worse with up-drift because of loss of the end-expiratory pause from the NE window (up-drift 10% mm-1, down-drift 7% mm-1, both P = .03). Scan efficiency also was reduced during sleep (to a nadir of 0%), secondary to loss of the end-expiratory pause, periodic breathing with oscillating end-expiratory position, and periods of apnea. The phantom model used actual diaphragm traces to evaluate the artifact resulting from diaphragm motion during acquisition. Artifact was considerably reduced by NE adaptive motion correction compared with NE gating alone (ghosting ratio 2.0 vs 2.8, P < .01). Artifact also was significantly reduced with up-drift if scan efficiency was maintained above 35% (P = .05). For optimal NE-gated MRCA, the following features are important: the NE window should be placed around the end-expiratory position; subjects should not sleep; scan efficiency should be monitored and the NE window should be repositioned if scan efficiency falls below 35%; and adaptive motion correction should be used.