This study evaluates prospectively whether baseline scores [Health Assessment Questionnaire (HAQ) and SF-36] can predict clinical and radiographic evolution in a cohort of early rheumatoid arthritis (RA) during a 3-year follow-up. Forty consecutive early RA patients were followed for 3 years, while receiving standardized treatment according to a pre-established protocol. HAQ and SF-36 were administered at the initial evaluation and at 3, 6, 12, 18, 24 and 36 months. Hands and feet radiographs were obtained at the initial evaluation and at 12, 24 and 36 months. Preselected outcomes were the occurrence of radiographic erosions, the achievement of an EULAR remission, low disease activity status and the need for biological therapy. The mean age at onset was 45 years with a 90% female predominance. Erosions were found in 42% of patients at T0 and in 70% after 3 years (P < 0.001). At T0, the proportion of patients in remission, low, moderate or high disease activity was 0, 0, 7.5 and 92.5% and 22.5, 7.5, 32.5 and 37.5%, respectively, at 3 years. The mean baseline HAQ score was 1.89 and 0.77 by the third year (P < 0.0001). Most SF-36 domains showed significant improvement except for general state and vitality. Biological therapy was deemed necessary in 22.5% of patients. The initial HAQ and SF-36 scores were not associated with clinical remission, bone erosions or the need for biological therapy at 36 months. The HAQ and SF-36 scores measured at baseline could not predict at 3 years, the preselected outcomes in a Brazilian cohort.