Aim: In the context of high-risk surgery, shared decision-making (SDM) is important but can be hindered by misalignment in expectations regarding post-operative outcomes between patients and clinicians. This study investigates the extent of this misalignment its’ impact on treatment choices, and its’ amenability to interventions that encourage perspective taking. Method: Lay participants with a Charlson co-morbidity index ≥4 and surgeons and anaesthetists were recruited. During an online experiment, participants forecast their expectations regarding short- and long-term outcomes of different treatment options for one of three hypothetical clinical scenarios: ischaemic heart disease, colorectal cancer, or osteoarthritis of the hip, and then choose between surgical or non-surgical treatment. Participants were asked to consider the scenarios from their own perspective (Estimation Task), and then to adopt the perspective of the other side (Perspective Task). Participants’ decisions (Surgery vs. non-surgical Alternative) were analysed using binomial generalized linear mixed models. Results: 55 lay-participants and 54 doctors completed the online experiment. Systematic misalignment in expectations between high-risk patients and doctors was observed, with patients expecting better surgical outcomes compared to clinicians. Patients forecast significantly higher likelihood of engaging in normal activities in the long-term (β=-1.09, SE=.20, t=-5.38, p<.001), lower likelihood of experiencing complications in the long-term (β=0.92, SE=.21, t=4.45, p<.001), and lower likelihood of experiencing depression in the in the short- and long-term (β=1.01, SE=.19, t=5.38, p<.001) after surgery than doctors. Patients forecast higher estimates of experiencing complications in the short-term when the non-surgical alternative was selected (β=-0.91, SE=.26, t=-3.50, p=.003) compared to doctors. Despite this misalignment, in both groups surgical treatment was strongly preferred (Estimation task: 88.7% of doctors and 80% of patients; Perspective task: 82.2% of doctors, 90.1% of patients). Conclusion: When high-risk surgery is discussed, a non-surgical option may be viewed as ‘doing nothing’, reducing the sense of agency and control. This biases the decision-making process, regardless of the expectations doctors and patients might have about the outcomes of surgery. Thus, to improve SDM and to increase patients’ agency and control over decisions about their care, we advocate framing the non-surgical treatment options in a way that emphasises action, agency, and change.