The objective of this study was to determine patterns among geriatric practitioners in prescribing agents that protect the gastrointestinal tract when nonsteroidal anti-inflammatory drug (NSAID) treatment is started for elderly patients. A questionnaire describing five scenarios of elderly patients requiring NSAID therapy asked respondents to choose gastrointestinal-protective agents for each scenario. Respondents were then asked to what extent four established risk factors for NSAID gastropathy (age, previous peptic ulcer, previous gastrointestinal bleeding, and heart disease) affected their choices. The choice of gastrointestinal-protective agent was compared with the training and experience of the respondents. This self-administered survey was provided to 821 randomly selected physicians from the membership of the American Geriatrics Society throughout the United States and Puerto Rico. Statistical Package for the Social Sciences (SPSS), version 6.1.4, was used to obtain frequencies. Of 821 surveys, 229 (28%) were returned. It was found that well elderly patients and nursing home residents were not treated with any gastrointestinal-protective agent by 64% (well elderly patients) and 32% (nursing home residents) of respondents. Among respondents who would prescribe, about half would choose misoprostol for a well elderly patient or a nursing home resident, whereas half or more preferred histamine H2-receptor antagonists. Twenty-three percent would not prescribe misoprostol when NSAID therapy was resumed after an active ulcer had healed, and 68% preferred H2 antagonists in that setting. The difference in response attributable to training/experience was less than 9%. Factors that did not affect prescribing patterns included the patient's age (15% to 62%) and heart disease (44% to 50%). The study concluded that age and heart disease are risk factors to which physicians give less consideration when choosing gastrointestinal-protective agents. Although misoprostol is the only agent approved by the Food and Drug Administration for prophylaxis against NSAID gastropathy, 23% of respondents chose not to prescribe misoprostol when NSAID therapy was resumed after an active ulcer had healed. Histamine H2-receptor antagonists were preferred over misoprostol for well elderly patients and nursing home residents. Training and experience were not responsible for differences among respondents' prescribing patterns.