We performed fasting and postprandial recordings of antroduodenal manometry in 21 normal volunteers, 13 patients with insulin-dependent diabetes mellitus and gastrointestinal symptoms, and 11 patients with the irritable bowel syndrome. None of the patients or volunteers had previously undergone an intestinal intubation study. Recordings could not be obtained from four of the diabetic patients due to failure to intubate the pylorus. Catheter migration led to incomplete antral data in a further 21% of all recordings. Due to the wide variations demonstrated by the normal volunteers, parameters of either the migrating motor complex (MMC) or the fed response could not differentiate between either of the patient groups and/or the controls. Similarly, while abnormal patterns of either fasting or postprandial motility were common in the diabetic patients, manometry had a sensitivity of only 67% in comparison to the less invasive radionuclide gastric emptying study. Furthermore, manometry failed to identify any diagnostic abnormality in irritable bowel patients; in particular, the incidence of "clustered" contractions was similar in all three groups. We conclude that short duration antroduodenal manometry is of limited diagnostic usefulness due to the difficulties in pyloric intubation in the presence of a dilated stomach and the intrinsic variability in normal motor patterns, perhaps excerbated by the stressful effects of the procedure itself in tube-naive subjects.