We reviewed 84 consecutive cases of peptic ulcer hemorrhage, which occurred, in an area of 270,000 people, from 1986 to 1988, in patients already hospitalized for other diseases (in-bleeders). These subjects were compared with a prospective series of 386 patients who initially bled as out-patients and were then admitted (out-bleeders). Of 84 hemorrhages in hospitalized patients, 41 followed major surgery, while 43 were associated with other severe conditions. Bleeding site was duodenal in two thirds. Mean age was 67 +/- 15 years versus 59 +/- 15 among out-bleeders. Fifty percent of in-bleeders had recently received nonsteroidal antiinflammatory drugs (NSAIDs), and one third were on anticoagulants and 10% on corticosteroids; in 39 (46%) bleeding was shown to be persistent or recurrent, 5 (5.9%) underwent endoscopic and 18 (21%) surgical therapy; 29 died (34%). The corresponding figures among out-bleeders were: further bleeding 80 (20.7%), endoscopic therapy 12 (3.1%), surgery 25 (6.5%), deaths 17 (4.4%). As regards in-bleeders, only active bleeding and endoscopic stigmata emerged as statistically significant risk factors for further bleeding. The latter was shown to be significantly related to mortality. The most relevant finding was, however, that NSAIDs and anticoagulants, in association with stress and aging, are very frequently involved in peptic ulcer bleeding of hospitalized patients. The fatal outcome of one third, despite all available treatments, highlights the importance of prevention against drug- and stress-related mucosal damage in in-patients suffering from severe diseases.