The effect of five approved tumour necrosis factor inhibitors (TNFi: infliximab, etanercept, adalimumab, certolizumab, and golimumab) on joint destruction in rheumatoid arthritis (RA) have been compared versus methotrexate (MTX) in randomized controlled trials (RCTs) but have not been compared directly to each other or to an otherwise untreated placebo control. The present analysis compares effects of standard doses, high doses, and low doses of TNFis on radiographic joint destruction in RA and relate these effects to MTX and placebo by means of a Bayesian network meta-analysis. We identified 31 RCTs of the effect of TNFis on joint destruction and 5 RCTs with controls, which indirectly could link otherwise untreated placebo controls to the TNFi treatments in the network. The previously untested comparison with placebo was performed to estimate not only the effect relative to another drug, but also the absolute attainable effect. Compared to placebo there was a highly significant inhibitory effect on joint destruction of infliximab, etanercept, adalimumab, certolizumab, and golimumab, which was about 0.9% per year as monotherapy and about 1.2% per year when combined with MTX. Although significantly better than MTX and placebo, golimumab seemed inferior to the remaining TNFis. There was no difference between original reference drugs (Remicade, Enbrel) and the almost identical copy drugs (biosimilars).