Background: Human papillomavirus (HPV) is a family of small double-stranded DNA viruses. It is associated with a series of cancer types, such as cervical, oropharyngeal, anal, vulvar, vaginal, and penile cancers. In the world, nearly 70% of HPV-related cancer cases are HPV types 16 and 18. In this study, we investigated HPV infection prevalence among a cohort of Chinese solid tumor patients. Methods: The study enrolled thousands of Chinese patients with different types of cancers. We finally collected 4,606 tumor tissue DNA (ttDNA), 7,296 plasma cell-free DNA (cfDNA), and 7,509 peripheral blood mononuclear cell genomic DNA (gDNA) samples, respectively. Using a targeted capture technique that covering the whole reference genome of HPV16 and HPV18 and the next generation sequencing (NGS) technology, we calculated the percentage of HPV16 and HPV 18 reads in all sequencing reads for each sample. Finally, we explored the prevalence of HPV infection in each cancer type. Results: Based on different types of DNA samples, the HPV infection rate was 38.75%, 28.62%, and 10.43% in ttDNA, cfDNA, and gDNA samples, respectively. Besides, the magnitudes of the mean percentage of HPV reads in all the sequencing reads in ttDNA, cfDNA and gDNA samples were 10^-5, 10^-6 and 10^-8, respectively, while in the HPV infection positive the percentages of ttDNA, cfDNA, and gDNA samples increased to 10^-4, 10^-5, and 10^-7, respectively. The prevalence of HPV infection rate in different cancer types was shown in table. Conclusions: ttDNA and cfDNA are better material for HPV infective detection compared with gDNA, and NGS is capable of HPV detection.