The predictive roles of symptom combination traditionally evaluated in traditional Chinese medicine (TCM) in the treatment of rheumatoid arthritis (RA) were explored. Three hundred and ninety six patients were randomly divided into 197 subjects receiving Western medicine therapy (WM) and 199 subjects receiving TCM therapy (TCM). A complete physical examination and 18 clinical manifestations typically assessed in TCM were recorded before the randomization. The ACR responses were used for efficacy evaluation. ACR20 and 50 responses with WM treatment were higher than in the TCM group. The 18 symptoms in RA could be clustered into 4 symptom combinations with factor analysis, which represent joint symptoms, cold pattern, deficiency pattern and hot pattern in TCM respectively. TCM would be more effective in patients with weak-symptom combination 3 (deficiency pattern in TCM), and WM would be more effective in patients with symptom combination 2 (cold pattern in TCM). Symptom combinations judged with TCM may have influence on the efficacy of therapy in the treatment of RA.