INTRODUCTION Disease-modifying anti-rheumatic drugs (DMARDs) currently form the mainstay of treatment of rheumatoid arthritis (RA). We aimed to evaluate the retention rates of "therapeutic segments" of DMARDs in patients with RA. METHODS This was a cross-sectional study of RA patients with at least one year of follow-up. A therapeutic segment is said to begin when one DMARD combination is instituted and it ends with a subsequent change. The disability index for each patient was calculated using a modified health assessment questionnaire. Retention rates were calculated using the Kaplan Meier survival analysis. RESULTS 375 DMARD courses in 102 patients were analysed. 99 courses were being continued at the time of the study and hence were censored for the purposes of analysis. The respective median (interquartile range [IQR]) retention period for segments containing methotrexate (MTX), sulfasalazine, hydroxychloroquine and leflunomide was 28 (15-45), 12 (3-20), 18 (9-24), 15 (4-32) months. The log-rank statistical test indicated that MTX was retained longer singly (median [IQR] 43 [32-70] months) than in combination (median [IQR] 19 [10-24] months) (p-value is 0.001). The commonest reason for the discontinuation of the DMARD segment was the disease "slipping out" of control (51.1 percent) followed by adverse effects (24.3 percent). Treatment termination on account of disease control was encountered in 16.3 percent of courses only. As many as 63 percent of single DMARD segments were changed because of disease "slip out" as compared to 41 percent of combination DMARD segments. Adverse effects were a more frequent cause of termination of the combination segments (32 vs. 15 percent). CONCLUSION MTX, used singly, had the highest retention rates among all the DMARDs used in RA patients. Disease "slip out" and adverse effects frequently required a change of the therapeutic segment.