Background: HPV genomic DNA in plasma and saliva has been widely studied, however more recently, circulating tumor human papillomavirus DNA (ctHPVDNA) has emerged as a reliable biomarker for surveillance in HPV+ oropharyngeal squamous cell carcinoma (OPSCC). A commercial assay for this biomarker distinguishes tumor-derived viral DNA (tumor-tissue modified viral DNA or TTMV) from other non-cancer associated sources of HPV DNA. The use of this technology has been previously described in plasma, but its utility in saliva is currently unknown. Methods: A prospectively collected and banked biospecimen repository was used to identify 46 patients with HPV+ OPSCC with paired pre treatment plasma and saliva samples. All samples were assessed for DNA integrity and TTMV using a clinically validated ddPCR-based assay (NavDx™; Naveris Inc, Natick, MA) to measure TTMV for HPV-16, -18, -31, -33 and -35 from frozen plasma and saliva samples. Retrospective chart review was performed to collect clinical and pathological data. Graphpad was used for statistical analysis. Spearman’s r was used to correlate TTMV copies in saliva and plasma. Wilcoxon test was used to compare between sample types. Mann-Whitney test was used for categorical variables. Results: TTMV DNA was detectable in 43 of 46 plasma samples and in 44 of 46 saliva samples. One plasma sample failed quality control measures, one of each sample type had undetectable TTMV, and one of each type was indeterminate. Of 41 evaluable patients with paired samples, there were 38 (93%) males, 36 (88%) were stage I-II, 5 (12%) were stage III-IV (AJCC 8^th, clinical staging), and 25 (61%) had a history of smoking with a median of 37.5 pack years. TTMV was significantly enriched in saliva compared to plasma (p<0.0001), with median copy number 14,139 copies/ml (IQR=193,339.5) and 774.7 copies/ml (IQR=4,826.1), respectively. There was a significant positive correlation between plasma and saliva TTMV levels (r=0.344, p=0.028). There was no difference in overall stage for either specimen type. There was a trend in both sample types toward higher TTMV in patients with a history of smoking. Pack-year history was available for 38 (93%) patients in the final cohort. When grouping by pack-years, plasma TTMV approached significance (p=0.058) while high saliva TTMV was significantly associated with >10 pack-year history (p=0.011). Conclusions: This is the first study to demonstrate successful quantification of tumor-tissue modified HPV DNA in saliva. Compared to plasma, pre treatment saliva samples demonstrated significantly higher levels of TTMV. TTMV distinguishes ctHPVDNA from other sources of HPV. These data highlight the potential use of TTMV detection in saliva for early detection of HPV+ OPSCC as well as its potential role in local surveillance after treatment. More research is needed to elucidate the effects of smoking on TTMV levels.