Measuring shared decision making (SDM) in clinical practice is important to improve the quality of health care. Measurement can be done by trained observers and by people participating in the clinical encounter, namely, patients. This study aimed to describe the correlations between patients' and observers' ratings of SDM using 2 validated and 2 nonvalidated SDM measures in clinical consultations. In this cross-sectional study, we recruited 238 complete dyads of health professionals and patients in 5 university-affiliated family medicine clinics in Canada. Participants completed self-administered questionnaires before and after audio-recorded medical consultations. Observers rated the occurrence of SDM during medical consultations using both the validated OPTION-5 (the 5-item "observing patient involvement" score) and binary questions on risk communication and values clarification (RCVC-observer). Patients rated SDM using both the 9-item Shared Decision-Making Questionnaire (SDM-Q9) and binary questions on risk communication and values clarification (RCVC-patient). Agreement was low between observers' and patients' ratings of SDM using validated OPTION-5 and SDM-Q9, respectively (ρ = 0.07; P = 0.38). Observers' ratings using RCVC-observer were correlated to patients' ratings using either SDM-Q9 (rpb = -0.16; P = 0.01) or RCVC-patients (rpb = 0.24; P = 0.03). Observers' OPTION-5 scores and patients' ratings using RCVC-questions were moderately correlated (rφ = 0.33; P = 0.04). There was moderate to no alignment between observers' and patients' ratings of SDM using both validated and nonvalidated measures. This lack of strong correlation emphasizes that observer and patient perspectives are not interchangeable. When assessing the presence, absence, or extent of SDM, it is important to clearly state whose perspectives are reflected.