I will present a history of the diagnostic practices associated with heart failure, and use this history to show that many philosophical assumptions made by physicians are problematic. Historical work shows that the development of diagnostic practices associated with heart failure was a complex, iterative and historically contingent process. This process has produced several logically incompatible diagnostic practices which select different groups of patients as diseased, some of which are potentially useful today in different medical contexts. Despite this, many cardiologists express their concern that physicians in general practice frequently misdiagnose the disease. Physicians appear to expect that there is a single ‘correct' method of diagnosis, which will classify patients optimally for any medical purpose. Cardiologists argue that physicians in general practice rely too much on the clinical signs and symptoms associated with the disease to make a diagnosis, and that echocardiographic measurements should be made as well. Some physicians, however, argue that heart failure is diagnosed accurately using symptoms and signs alone. All parties use empirical evidence to support their arguments. However, their arguments are problematic, and often self-contradictory. I will present an analysis of some of these arguments, pointing out how they are problematic, where they are self-contradictory, and how they are revealing of philosophical assumptions made by physicians. I will argue that the contradictions identified are the result of physicians' attempts to present these different practices as part of a single ‘correct' version of heart failure. Overall I will argue that integrated historical and philosophical research can be used to inform medical practice.