Comparing outcomes of different therapeutic interventions is a vital part of progress in rheumatology, and considerable advances have been made in recent years toward consensus in outcome assessment. The conflicting results of longitudinal and cross-sectional studies of rheumatoid arthritis (RA) have been due, at least in part, to variable study designs. The few reports of prospective studies of recent-onset RA have been limited by the paucity of numbers at follow-up. Information on outcome and prognostic factors in RA requires prospective studies on large numbers of patients recruited from an early stage. The practising clinician needs easily applied prognostic factors for the effective management of patients with RA. Predicting erosive disease has been more successful than predicting functional outcome, but neither, as yet, have the accuracy required in early disease for the use of single or combined powerful, and potentially more toxic, second-line (disease-modifying) therapies. The Early Rheumatoid Arthritis Study (ERAS) was formed to study these aspects in more detail with large numbers of patients. Initial findings have shown that 15% of patients went into sustained clinical remission early, response to the first disease-controlling drug was satisfactory in 70% of patients and functional measures at presentation were by far the most predictive variables for physical disability at 3 yr.