This paper critically evaluates central themes of values-based medicine (VBM). First, I discuss the 'non-descriptivist' conception of value judgments at the heart of VBM. According to it, no inferences can rationally be drawn from factual criteria to value judgments and the inferences that are naturally formed are a matter of human psychology. I argue, however, that it is an essential feature of value judgments that they are themselves subject to normative assessment. This implies an important role for an evaluatively rich notion of psychology. Second, I discuss the role of 'dissensus' in VBM and argue that, although value judgments need not converge, the processes for exploring conflicting values must, themselves, be normatively assessable. Third, I examine how these aspects interact in complex real cases such as the case study of Diane Abbot.