Alanine aminotransferase (ALT) and ultrasound (US) are the most commonly used tools for detecting non-alcoholic fatty liver disease (NAFLD). No direct comparison of these two modalities in children exists. We aimed to compare head-to-head the diagnostic accuracy of ALT and US and their combination for detecting NAFLD in children with obesity. Ninety-nine children with severe obesity underwent simultaneous serum-ALT and abdominal ultrasound (US steatosis score 0–3). Proton magnetic resonance spectroscopy was used as reference standard for detecting steatosis/NAFLD. ROC curve analyses were performed to determine diagnostic performance and to determine optimum screening cut-points aiming for a specificity ≥ 80%. The area under the ROC (AUROC) of ALT and US were not significantly different (0.74 and 0.70, respectively). At the optimal ALT threshold (≥40 IU/L), sensitivity was 44% and specificity was 89%. At the optimal US steatosis score (≥ 2), sensitivity was 51% and specificity was 80%. Combining ALT and US did not result in better accuracy than ALT or US alone. Conclusion: ALT and US have comparable and only moderate diagnostic accuracy for detecting hepatic steatosis in children with obesity. A stepwise screening strategy combining both methods does not improve diagnostic accuracy.What is Known:• Alanine aminotransferase (ALT) and ultrasound (US) are the most commonly used tools for detecting non-alcoholic fatty liver disease (NAFLD).• ALT and ultrasound have mediocre accuracy in detecting steatosis in children with obesity.What is New:• In a head-to-head comparison, the difference in diagnostic accuracy of ALT and ultrasound in detecting steatosis is not significant.• A stepwise screening strategy combining both methods does not improve diagnostic accuracy.