Background: Cross-sectional gastric antral area (GAA) measurements by ultrasonography (US) have been proposed for preoperative assessment of gastric volume in adults but not been validated in children. This study investigates whether in children gastric volumes can be predicted by US performed in different patient positions. Method: Gastric fluid and air volumes were examined by magnetic resonance imaging before or up to 120 min after ingestion of 7 ml·kg(-1) diluted raspberry syrup in healthy volunteers who had fasted overnight. GAA was measured with US three times each in supine (SUP), elevated 45° degree supine (E45) and right decubital (RDC) position using imaging planes defined by vascular landmarks. Correlation coefficients (Pearson) between GAA and gastric volumes were calculated and Bland-Altman analysis performed. Results: Sixteen children aged from 6.4 to 12.8 (9.2) years were included in 23 examinations: 6 after overnight fasting, 3 directly after, and 14 with a delay of 74 ± 35 min after fluid intake. GAA was 221 ± 116, 218 ± 112, and 347 ± 188 mm(2) for SUP, E45, and RDC position, respectively. The best correlation between body weight corrected total gastric/gastric fluid volume (TGV(w) /GFV(w) ) with GAA was found for RDC position (R = 0.79; P < 0.01/R = 0.78; P < 0.01). Bias and precision of calculated and measured GFV(w) was 0 ± 2.8 ml·kg(-1) . Conclusion: Correlations between GAA and TGV(w) or GFV(w) in children are best in the RDC position, but not sufficient to predict GFV(w) with a given GAA. Interpretation of isolated GAA values may be misleading.