To evaluate whether circulating fibroblast activation protein a (FAPa) could serve as a biomarker for the diagnosis of esophageal squamous cell carcinoma (ESCC), enzyme-linked immunosorbent assay (ELISA) was used to detect plasma FAPa in 556 participants including ESCC group, benign esophageal disease group, healthy controls and other cancer controls group. The levels of plasma FAPa were significantly decreased in ESCC patients (P < 0.001) and showed a positive correlation with HDL-C levels (R = 0.372, P < 0.001). The sensitivity and specificity of plasma FAPa were 56.1% and 85.6% based on the optimal cut-off (49.04 ng/ml, AUC = 0.714). The combination of FAPa and the traditional biomarkers (CEA, CYFR211 and SCCA) improved the sensitivity (41.5%) without compromising the specificity (95.0%). Contradictorily, the immunohistochemical staining revealed the overexpression of FAPa in stroma of ESCC tissues. So the source of soluble FAPa was further explored by qRT-PCR, Western blotting, ELISA and immunoprecipitation in fibroblast cell lines and mouse xenograft models. We found that the plasma FAPa was not correlated with the FAPa expressed in tumor, and the multi-organ might contribute to the circulating levels of FAPa including skeletal muscle, liver and bone marrow. These results indicated that the low plasma FAPa level might due to the systemic reaction to the presence of tumor and circulating FAPa level might be a potential indicator for diagnosing ESCC.