Magnetic resonance imaging has been recently proposed as a promising, noninvasive technique to assess the motility of the gastric antrum. However, so far the reproducibility and dependence on test meal composition has not been evaluated. In this study, snapshot echo-planar magnetic resonance imaging was used to measure the frequency, propagation speed and percentage occlusion of antral contractions in 28 healthy volunteers. They were fed either liquid (n=12), mixed liquid/solid (n=8) or mixed viscous/solid (n=8) nutrient (1350 kJ) test meals, and a total of 208 motility measurements were performed. No effect of meal type on antral motility parameters was observed. Antral contraction frequency was 3.0 +/- 0.2 min(-1) (mean +/- SD, n=164), propagation speed was 1.6 +/- 0.2 mm s(-1) (n=164) and the percentage occlusion was 58 +/- 14% (n=76). Overall, 21% of measurements did not provide useful antral motility data, because, in the supine position, the antrum was not filled by the test meal. Simple methods to overcome this and reduce scanning time to a minimum are proposed. The results show that the noninvasive magnetic resonance imaging evaluation of antral motility is accurate and reproducible and has potential to become a standard tool for such investigations.