To study the symptoms of NSAID-associated gastroduodenal bleeding, 94 patients (median age 71 years, range 19-90), were included in a prospective, clinical trial where hematemesis or melena from gastroduodenal ulceration or haemorrhagic/erosive gastritis were the inclusion criteria. NSAID use within one month was studied in relation to subjective symptoms prior to admission and to clinical course of the episode. Significantly fewer of the NSAID users (n = 54) than the non-users (n = 40) had experienced prior peptic ulceration or dyspeptic symptoms. Otherwise, no differences were seen between users and non-users, as regards pre-admission epigastric pain, heartburn or nausea. Also, the clinical course was similar in the two groups. We also found sporadic and regular NSAID use to be similar in this respect. These data do not support the alleged masking of ulcer symptoms by NSAIDs in bleeding ulcers.