Radionuclide studies have gained wide acceptance in the evaluation of infants and children with gastroesophageal reflux (GER). For correct interpretation of scan results, knowledge of inter-observer and intra-observer variability and minimum detectable reflux volume is essential. In this study, we evaluated the methodological issues underlying the visual assessment of GER and time activity curve analysis. An in vitro model of stomach and oesophagus was established to determine the minimum detectable reflux by placing various volumes and concentrations representing the diluted activity in the stomach. In the clinical part 99 patients were imaged for 1 h after oral administration of Tc-99m sulfur colloid. Eleven patients were excluded from the study either due to incomplete clinical data or suboptimal image quality. Frames of 16 s each, and time-activity curves which were generated after drawing regions of interest from the oesophagus, were read three times by an experienced nuclear medicine physician and a resident in training. On the phantom study, the concentration, volume and duration were the determining factor for the visualization of reflux. In the clinical part, the overall incidence of GER in 88 patients was 69%. The mean intra- and inter-observer reproducibility (kappa values) was 0.76 and 0.7065, respectively. Agreement was slightly higher in the analysis of time-activity curves (0.767 and 0.731). Our results indicate that GER may be reproducibly analysed on scintigraphy by the same and different observers with varying levels of training. Its visualization is associated with reflux duration, volume and dilution factor of radioactivity. ((C) 2003 Lippincott Williams Wilkins).