Thirteen infants who had undergone 24 hour oesophageal pH monitoring to diagnose gastro-oesophageal reflux had a second study carried out to see if the results were reproducible. The studies were done without restricting the babies' activities. Appreciable differences were found, the percentage of the total time during which the pH was less than 4 varying by up to 3.7-fold between the two tests. The differences were largely the result of biological rather than technical variability. From these results estimates were made of the reliability of a single diagnostic study and the size of changes that would be necessary to show the effect of treatment. These findings have a considerable impact on the diagnosis of abnormal gastro-oesophageal reflux and its response to treatment whether using 24 hour pH monitoring or any other method of measurement.