Physicians may choose one of several strategies when initially uncertain about making a specific therapeutic recommendation. The authors investigated how patients' satisfaction is affected by disclosure of uncertainty and its attempted resolution during a clinical encounter. Three hundred and four patients awaiting appointments at a university hospital's ambulatory medical clinic were randomized to view one of five videotapes (VTs) of a patient seeking advice about antimicrobial prophylaxis for a heart murmur. In VT-1 and VT-2, the physician disclosed no uncertainty and prescribed therapy. In VT-3, VT-4, and VT-5, the physician openly conveyed uncertainty but then: (VT-3) prescribed antibiotics without resolving his uncertainty; (VT-4) consulted a reference book with the patient present, then prescribed; or (VT-5) checked a computer with the patient present, then prescribed. Patients rated their satisfaction with the physician on a standardized questionnaire. Differences in satisfaction between the five VTs were significant (p = 0.001), with the highest ratings found for VT-1 and VT-2, where no uncertainty was disclosed. The lowest ratings in satisfaction were found when the physician expressed but then ignored uncertainty (VT-3) or examined a textbook (VT-4). Global satisfaction was inversely and significantly correlated (r = -0.47) with the patients' perception of uncertainty in the physician. The manner in which clinical uncertainty is disclosed to patients and then resolved by the physician appears to affect patients' satisfaction.