Over 75% of high-grade serous ovarian cancer (HGSOC) patients are found to have advanced disease at initial diagnosis, frequently presenting with peritoneal metastasis and ascites which are considered further promoting the metastasis of HGSOC. The proteins in fluid portion of ascites are mainly exsit in free style, small and large extracellular vesicles and the composition of them needs to be further explored for characterization. In this study, we quantitatively analyzed 5650 proteins in 39 samples from 13 patients with HGSOC using data-independent analysis (DIA) mass spectrometry, focusing on liquid ascites, small extracellular vesicles (sEVs) and large extracellular vesicles (lEVs). Our results revealed distinct proteomic profiles among the three categories and identified eight expression patterns through trend clustering analysis of the differential proteins. Through the integration of single-cell RNA-seq data of ascites cells, we discovered that proteins significantly enriched in sEVs were more frequently derived from fibroblasts rather than malignant cells, a pattern not observed in lEVs. Subsequently, we focused on a set of 242 proteins with expression trends of sEV>lEV>liquid ascites. To further analyze their correlation in metastasis and prognosis, we investigated plasticity-related genes in transcriptome datasets from GEO and MSigDB database, and performed prognostic analyses using data from GEO, EGA, and TCGA database, identifying four genes closely linked to tumor cell plasticity and patient survival. In conclusion, we comprehensively investigated the protein composition of HGSOC ascites, identifying proteins exhibiting enrichment in sEVs and are associated with tumor plasticity and prognosis. This offers a significant complementary to traditional investigations centered on patient ovarian cancer cells, deepening our understanding of the disease.