Background: Computational designing of novel proteins that can bind to a specific target protein to modulate its functions is a promising and challenging approach in synthetical biology and therapeutical protein design.Instead of designing a complicated new structure, short peptides with only one single helix can be used for some systems.Methods: We have developed a general method for designing helical peptides that can bind with a target protein.First, appropriate helix binding sites on the target protein surface are identified based on simple geometry matching and the buried surface hydrophobicity.Second, for a binding site, initial poly-alanine helices are designed.Then, protein sequences and binding conformations are evolved in silico by a de novo protein design strategy, for example, the protein design program, Rosetta can be used.The computer-generated sequences were screened by scoring functions which include estimation of binding affinity, folding stability, and aggregation, etc.Results: This method has been applied to design antagonist for tumor necrosis factor (TNF).Several single-helix binding sites on TNF at the TNF and receptor binding interface were identified, and the best one, for which the corresponding binding helices are along with the β-strands in the TNF receptor, was selected for the subsequent design.The designed peptides were expressed as GST fusion protein or chemically synthesized and tested by surface plasmon binding assay and cell based assay.Active peptides were identified and their binding pose was verified by mutation study.Conclusions: We have developed a computational strategy for design α-helical peptides that can specifically bind to a protein target.and successfully applied it in designing a novel peptide that can bind with TNF and inhibit its activity.This method could easily be used to other cases to identify helix binding sites in proteins and de novo design binding peptides for the target .