We compare two clinical series of patients with fever of unknown origin (FUO) during two different periods of time (1982-1989 and 1997-2004) in the department of Internal Medicine at La Paz University Hospital, Madrid. We have analysed the changes in aetiology and diagnostic methods in these separated periods. We describe a decrease in the incidence of this clinical entity which we think to be due to the improvement of the diagnostic methods. This provides an earlier diagnosis and so patients do not reach criteria for classical FUO. Due to these improvements we have also assessed a change in the final diagnoses and the means used to reach them which, in general, have been less interventionist.