Treatments that block tumour necrosis factor (TNF) have major beneficial effects in several autoimmune and rheumatic diseases, including rheumatoid arthritis. However, some patients do not respond to TNF inhibitor treatment and rare occurrences of paradoxical disease exacerbation have been reported. These limitations on the clinical efficacy of TNF inhibitors can be explained by the differences between TNF receptor 1 (TNFR1) and TNFR2 signalling and by the diverse effects of TNF on multiple immune cells, including FOXP3⁺ regulatory T cells. This basic knowledge sheds light on the consequences of TNF inhibitor therapies on regulatory T cells in treated patients and on the limitations of such treatment in the control of diseases with an autoimmune component. Accordingly, the next generation of drugs targeting TNF is likely to be based on agents that selectively block the binding of TNF to TNFR1 and on TNFR2 agonists. These approaches could improve the treatment of rheumatic diseases in the future.