Crohn's disease is a serious and debilitating gastrointestinal disorder with a high, unmet need for new treatments. Biologic agents have the potential to alter the natural course of Crohn's disease but present known risks of potential serious adverse events (SAEs). Previous studies have shown that patients are willing to accept elevated SAE risks in exchange for clinical efficacy. Gastroenterologists and patients may have similar goals of maximizing treatment benefit while minimizing risk; however, gastroenterologists may assess the relative importance of risk differently than patients. To (a) understand how gastroenterologists caring for patients with Crohn's disease balance benefits and risks in their treatment decisions and (b) compare their treatment preferences with those of adult patients with Crohn's disease. Both patient and gastroenterologist treatment preferences were elicited using a web-based, choice-format conjoint survey instrument. The conjoint questions required subjects to choose between 2 hypothetical treatment options with differing levels of treatment attributes. Patients evaluated the treatment options for themselves, and gastroenterologists evaluated the treatment options for each of 3 hypothetical patient types: (a) female aged 25 years with no history of Crohn's disease surgery (young), (b) male aged 45 years with 1 Crohn's disease surgery (middleaged), and (c) female older than 70 years with 4 Crohn's disease surgeries (older). Treatment attributes represented the expected outcomes of treatment: severity of daily symptoms, frequency of flare-ups, serious disease complications, oral steroid use, and the risks of 3 potentially fatal SAEs - lymphoma, serious or opportunistic infections, and progressive multifocal leukoencephalopathy (PML) - during 10 years of treatment. Maximum acceptable risk (MAR), defined as the highest level of SAE risk that subjects would accept in return for a given improvement in efficacy (i.e., the increase in treatment risk that exactly offsets the hypothetical increase in treatment benefit), was calculated using preference weights (parameter marginal log odds ratios) that were estimated with conjoint analysis (random parameters logit models). Gastroenterologists' and patients' mean MARs for 3 SAE risks were calculated for 6 improvements in Crohn's disease symptoms, and gastroenterologists' preference weights for each of the 3 patient profiles were compared. Gastroenterologists' MARs for a hypothetical middle-aged patient were then compared with predicted MARs derived using data from the patient study for male patients aged 40 to 50 years with 1 surgery. After exclusion of nonrespondents (n = 4,021 of 4,422 gastroenterologists; n = 681 of 1,285 patients) and nonevaluable respondents (n = 86 gastroenterologists; n = 24 patients), 315 gastroenterologists and 580 patients were included in the final analytic samples. There were no statistically significant differences in gastroenterologists' preference weights for the middle-aged versus young patient profiles. However, preference weights indicated that gastroenterologists are more concerned about 5% side-effect risks for the older patient profile than for the middle-aged patient profile. For symptomatic improvements from severe symptoms to remission, gastroenterologists' highest MARs were for lymphoma: 6.21%, 8.99%, and 25.00% for the young, middle-aged, and older patient types, respectively. In analyses of improvements from severe to moderate symptoms and from moderate symptoms to remission for hypothetical middle-aged patients, gastroenterologists' 10-year risk tolerance ranged between 1.96% lymphoma risk in return for an improvement from moderate symptoms to remission and 4.93% lymphoma risk for an improvement from severe to moderate symptoms; patients' 10-year risk tolerance for middle-aged patients ranged between 1.52% PML risk in return for an improvement from severe to moderate symptoms and 5.86% infection risk for an improvement from moderate symptoms to remission. On average, gastroenterologists and patients disagreed about how much risk is tolerable for improvements in efficacy. In exchange for improvements from severe to moderate symptoms for the middle-aged patient profile, gastroenterologists were significantly more tolerant than patients of treatment risks of PML (P < 0.001) and serious infection (P = 0.001) but not lymphoma (P = 0.230). In contrast, in exchange for improvements from moderate symptoms to remission for the same patient profile, patients were significantly more tolerant than gastroenterologists of treatment risks for serious infection (P < 0.001) and lymphoma (P < 0.001) but not PML (P = 0.158). Gastroenterologists and patients have well-defined preferences among treatment attributes and are willing to accept tradeoffs between efficacy and treatment risks. However, risk tolerance varies depending on the type of patient for whom gastroenterologists are being asked to consider treatment. In rating treatment preferences for patients with a middle-aged profile, gastroenterologists are less tolerant of SAE risks than patients in exchange for improvement from moderate symptoms to remission.