Tumor necrosis factor is a major pro-inflammatory cytokine which triggers various physiological consequences by binding to and trimerizing its receptors, and has been the single most sought-after drug target for intervening autoimmune diseases such as rheumatoid arthritis and psoriasis. However, current TNF-α blockers, including soluble receptor-Fc fusion and therapeutic antibodies, are all dimeric in structure, whereas their target TNF-α itself is homotrimeric in nature. Here we describe the development of a trivalent soluble TNF receptor and show that it is a more potent than the dimeric TNF receptor decoys in inhibiting TNF-α signaling both in vitro and in vivo. The process involves gene fusion between a soluble receptor TNFRII with a ligand binding domain and a trimerization tag from the C-propeptide of human collagen (Trimer-Tag), which is capable of self-assembly into a covalently linked trimer. We show that the homotrimeric soluble TNF receptor (TNFRII-Trimer) produced with such method is more potent in ligand binding kinetics and cell based bioassays, as well as more efficacious in attenuating collagen-induced arthritis (CIA) in a mouse model than its dimeric TNFRII-Fc counterpart. Thus, this work demonstrates the proof of concept of Trimer-Tag and provides a new platform for rational designs of next generation biologic drugs.