Alternative tumor necrosis factor-${alpha}$ (TNF-${alpha}$) inhibitors and non-TNF biologics are available as treatment options for rheumatoid arthritis patients who exhibit inadequate response to TNF-${alpha}$ inhibitor (TNF-IR patients). These agents have considerable efficacy compared with placebo, but head-to-head comparisons among these agents have not been performed. The objective of this study was to use Bayesian approach to compare the effectiveness of cycling TNF-${alpha}$ inhibitors versus switching to non-TNF biologics in TNF-IR patients. A systematic review was conducted using MEDLINE and Cochrane library. Key endpoints were the American College of Rheumatology (ACR) responses of 20/50/70 and the health assessment questionnaire (HAQ) score change at six months. Bayesian outcomes were calculated as the probability that OR is greater than one and HAQ score change difference is less than zero. Compared with TNF-${alpha}$ inhibitors, non-TNF biologics were associated with higher ACR response rates; in ACR20, the OR was 1.639 for abatacept [P(OR>1) = 90.7 %], 1.871 for rituximab [P(OR>1) = 96.2 %] and 3.52 for tocilizumab [P(OR>1) = 99.9 %]. Similar trends were shown in the HAQ change comparison; the median differences (MD) were -0.259 for abatacept [P(MD<0) = 100 %], -0.160 for rituximab [P(MD) = 98.2 %], and -0.200 for tocilizumab [P(MD) = 99.3 %]. In conclusion, switching to non-TNF biologics was more effective than cycling TNF-${alpha}$ inhibitor in TNF-IR patients.