This doctorate thesis provides an insight into the biology and dynamics of circulating microRNAs, demonstrating their potential to become crucial biomarkers for a better surveillance and prognosis of cancer. MicroRNAs (miRNAs) are small, single-stranded non-coding RNAs, 19-25 nt long, with a key role in the post-transcriptional regulation of gene expression, repressing the translation of target mRNAs through partial base-pair complementarity with their 3'-UTR. They can be released in the extracellular medium, being protected from RNases by association with various transporters and reach body fluids and circulation, participating in intercellular communication. Their remarkable stability and manageable diversity in circulation, as well as the fact that they derive from both malignant and normal cells make them very attractive biomarker candidates, potentially reflecting tumor state and dynamics. We have been focusing on ovarian (OvCa) and nasopharyngeal (NPC) carcinomas, attempting to elucidate the relation between plasma miRNAs and the prognosis of OvCa after first-line treatment, as well as to evaluate their use in the detection of early response of NPC tumors to treatment. Serous epithelial ovarian carcinoma is the most frequent ovarian and the most aggressive gynecologic malignancy. The absence of early symptoms and the insufficiency of modern means to accurately map residual disease and assess treatment outcome highlight the need for new diagnostic and prognostic biomarkers. Using sequential plasma samples from OvCa patients before and after first-line treatment, we studied a pre-selection of miRNAs, comparing them to samples from benign pelvic lesions and healthy women. MiR-200b exhibited a distinct higher concentration in malignant samples before treatment compared to both non-cancerous groups. Pre- and post-treatment assessment of miR-200b in parallel with the standard biomarker CA125 revealed distinct variations and a significant correlation of miR-200b variation with the progression-free survival (PFS) of the patient. We suggest that miR-200b could eventually be used as a supplementary biomarker for estimation of the remission upon treatment completion. Nasopharyngeal carcinoma (NPC) on the other hand is a tumor consistently associated to latent Epstein-Barr virus (EBV) infection of the malignant cells, presenting a unique geographical incidence pattern. The deep position of the tumor makes it tough to access surgically, with biomarkers assessing different therapeutic approaches being greatly needed. Studying a new oral form of the demethylating agent 5-azacytidine, proven to be promising for one third of NPC patients receiving it as a monodrug, we attempted to identify impact of the drug on the expression of viral miRNAs and proteins. Despite the latent viral infection, viral miRNAs are abundantly expressed, attracting interest in EBV-associated malignancies. Treating four in vivo developed NPC tumor models for two weeks, we observed clear response in two of them, in a dose-dependent manner. Protein analysis showed an induction of the immediate-early lytic protein BZLF1, solidifying previous evidence of partial activation of the viral lytic cycle by 5-azacytidine. MiRNA analysis confirmed robust expression of BART and absence of BHRF1 miRNAs at baseline status of NPC. Upon treatment, we observed an induction of BHRF1 miRNAs in both tumor and plasma of treated mice. This induction was successfully validated in a following one-week treatment and completed by a recorder de novo expression of the BHRF1 mRNA, transcribed within the BHRF1 miRNA loci. A weaker induction of BHRF1 miRNAs was also recorded after treatment with standard chemotherapeutic agents, suggesting a potential clinical utility of these miRNAs as circulating biomarkers for detection of early response to treatment. We are further working to confirm this induction by chemotherapy and extend our study to plasma samples derived from treated patients.