The aim of the present study was to compare computer-based and visual/manual scoring of pressure recordings from the gastric antrum and duodenum. The computer system was based on three separate principles of recognition: moving average, tracking base line, and threshold filters. Computer scoring resulted in the detection of more contractions than visual/manual scoring, owing to inclusion of artefacts of duration and amplitude above the selected thresholds. Owing to its unreliability, visual/manual scoring led to inclusion of some contractions of below-borderline amplitude or duration. The median difference in the frequency of contractions was, however, 0 min-1. There was close agreement on amplitudes of contractions. As the differences between visual/manual analysis and computer-aided scoring by the present system are of an unimportant magnitude and as computer-based scoring is consistent, we conclude that computer-aided analysis is preferable to visual/manual analysis.