‘Establishing the causes of a disease’ makes the point that any association between an exposure and a disease that is reasonably well established in a study needs to go through a process of interpretation before a conclusion can be drawn about its nature, causal or non-causal. The following questions are asked: Did the exposure precede the disease? How strong is the association? Does the association become stronger with increasing exposure? Is the association consistent? Is the association specific? Is the association consistent with other biological evidence? Has the association any analogue? Is the association coherent across different studies?