We investigated the effect of long-term corticosteroid usage in suppressing the progression of functional disability in patients with early rheumatoid arthritis (RA). We studied 3,982 RA patients, who had continuous enrollment for at least 3 years, among 9,132 RA patients enrolled in an observational cohort study, IORRA, in Tokyo, Japan, from 2000 to 2007. The DAS28 and Japanese version of Health Assessment Questionnaire (J-HAQ) scores were collected at 6-month intervals (each phase). Among these patients, those with DAS28 values under 3.2 in all phases and RA disease duration under 2 years at study entry were selected as "early RA patients with well-controlled disease". These patients were further classified into 3 groups based on average months of steroid usage per year: Non-users, Medium-users, and Frequent-users. Multiple linear regression analysis was used to study the relationship between steroid usage and the final J-HAQ scores. Among the 3,982 patients, 109 had DAS28 values under 3.2 in all the phases and were selected as study cohort. The average Final J-HAQ in Non-user (N = 64), in Medium-user (N = 25), in Frequent-user group (N = 20) was 0.04, 0.06, and 0.33, respectively. Multiple linear regression analysis after adjusting for all potential covariates confirmed that frequent steroid usage was the most significant factor associated with higher final J-HAQ scores (P < 0.05). Frequent steroid usage was associated with significantly higher final J-HAQ scores in early RA patients, even though their disease was managed efficiently by maintaining the DAS28 values under 3.2 over a long-term period.