The invention belongs to the field of tumor purity detection, and especially relates to a model construction method for identifying a tumour purity sample and application thereof. The method comprisesthe steps of: based on targeted capture sequencing data of a tumor sample with known tumor purity, acquiring index data including a somatic cell copy number variation amplitude, heterozygote germlinemononucleotide variation and a somatic cell allele variation fraction; and associating the index data with the tumor purity of the tumor sample with known purity to construct an identification model.In the method, somatic mutation and germ cell variation are combined, the index data such as the somatic cell copy number variation amplitude, the heterozygote germline mononucleotide variation, themodel constructed by integrating the index data such as the somatic cell allele variation fraction and the like is used to solve the problem that the purity of tumors is difficult to identify due to few mutation sites in interval chip capture sequencing such as panel sequencing is solved, low-purity samples can be accurately identified, multiple cross confirmation among different indexes is achieved, and the reliability is high.