With clinical information it is frequently difficult or impossible to make objective measurements to prove that a particular finding is correct. In these circumstances, the agreement of experienced clinicians is the best guide to the accuracy of clinical findings. However, studies of clinical judgement usually show a distressingly high level of disagreement among clinicians. We examine why such disagreement is so common, how clinical disagreement can be measured and how clinicians can be trained so that they are able to agree on important findings.