Two questions were examined, namely, 1) whether pH 4 is really the optimal threshold for the definition of acid gastroesophageal reflux, and 2) to what extent shifting of the upper limits of normal affects sensitivity and specificity of 24-h pH monitoring. To answer these questions, we studied 74 patients with proven reflux disease and 37 asymptomatic volunteers, using ambulatory 24-h esophageal pH monitoring. Gastroesophageal reflux was defined as episodes with esophageal pH of less than the threshold values 3.0, 3.5, 4.0, 4.5, or 5.0, respectively. For each of these pH thresholds, the percentage time with esophageal pH below the threshold was calculated separately for periods of upright and supine body position. Two-dimensional receiver-operating-characteristic (ROC) analysis was used to define upper limits of normal. A maximum of sensitivity, specificity, and rate of correct decisions (all 89%) was obtained using pH 4 for the definition of gastroesophageal reflux, although other pH thresholds were not much worse. On the basis of pH 4, the upper limits of normal could be shifted around the "optimal upper limit of normal" within a certain limit without considerable loss of accuracy of pH monitoring. This may explain the divergences between upper limits of normal obtained by different laboratories. In conclusion, 1) the threshold pH 4 should further be used for the definition of acid gastroesophageal reflux, and, 2) within certain limits, shifting of the upper limits of normal has little effect on the accuracy of pH monitoring in gastroesophageal reflux disease.