The Oslo experience with early rheumatoid arthritis (RA) is based on studies performed within the setting of a register of RA patients and of the longitudinal follow-up of patients with early RA. This article discusses some relevant issues for research on early arthritis: whether the current RA classification criteria are appropriate, the shift in incidence toward elderly patients, and the heterogeneity of the disease. Our data clearly show that 3 of the items of the classification criteria are infrequently fulfilled early in the disease and that RA most frequently starts after the age of 60. Our data also suggest that disease onset may be defined either as symptom onset or as when the classification criteria are fulfilled. The choice between these 2 options does not seem to influence the longitudinal results.