Should the assessment of decision‐making capacity (DMC) be risk sensitive, that is, should the threshold for DMC vary with risk? The debate over this question is now nearly five decades old. To many, the idea that DMC assessments should be risk sensitive is intuitive and commonsense. To others, the idea is paternalistic or incoherent, or both; they argue that the riskiness of a given decision should increase the epistemic scrutiny in the evaluation of DMC, not increase the threshold for DMC. We respond to the critics’ main concerns by providing a comprehensive account of how risk‐sensitive DMC is coherent, avoids paternalism, and best fulfills the epistemic goal of DMC evaluations.