There are considerable research and health service benefits in understanding current trends in a disease as common and as severe as rheumatoid arthritis. Unfortunately in the absence of both a high and constant case fatality rate and a population based morbidity registration system, there is no direct means of assessing changes in incidence in contrast, for example, to the situation with cancer. All the available data are inadequate and subject to considerable bias. It is of note however that all the available data are relatively consistent in pointing to a recent reduction in incidence over the last two decades coupled with a decrease in case severity. The immediate problem remains therefore to set up a true population based recording system which will permit the identification of trends both of incidence and severity. It should not be necessary for this to be established nationwide and a study area within a country could be chosen providing its population was representative and of sufficient size. For the latter a minimum seize of 500,000 persons would be needed.