Mercury (Hg) is a naturally occurring metal whose concentrations have been greatly amplified by human activities, notably artisanal gold mining and coal combustion. This global pollutant is transported through the atmosphere, soils, and aquatic environments, where it undergoes numerous biogeochemical transformations. Among these, the formation of monomethylmercury (MMHg) is of particular concern: this neurotoxic compound accumulates and biomagnifies along marine food webs, exposing humans through seafood consumption. Despite the Minamata Convention (2013), understanding of the marine mercury cycle remains limited due to ultra-trace concentrations (<1 ng.L⁻¹) and the high cost of conventional discrete sampling. This thesis evaluates the potential of passive samplers based on Diffusive Gradients in Thin Films (DGT) for monitoring MMHg and in organic mercury (HgII) in marine waters. DGT devices integrate concentrations over days to weeks, providing time-weighted averages more representative of geochemical background than discrete sampling. The main objective is to overcome analytical constraints limiting DGT application to mercury monitoring. An optimized elution protocol was developed for commercial DGTs equipped with 3-mercaptopropyl-functionalized silica (3MFS) resin. It achieved near-quantitative MMHg recovery (94 % ± 3 %) through three successive ultrasonic extractions with acidic thiourea (5 mM). Polyacrylamide binding gels significantly improved efficiency and reproducibility compared to agarose gels. The protocol was validated during two field campaigns: (i) in Peruvian coastal waters, revealing vertical MMHg gradients linked to benthic production, and (ii) in the open ocean (Southwest Pacific), where monthly deployments over a year with an automated system highlighted potential seasonal variability in MMHg concentrations. Two approaches were investigated for HgII. The first, based on AG1-X4 anion-exchange resin, showed selectivity but slow accumulation, requiring further optimization. The second combined dual analysis of 3MFS DGTs: (a) gel elution followed by MMHg and HgII determination by PT-GC-AFS (purge, trap, gas chromatography, atomic fluorescence spectrometry), and (b) analysis of residual HgII by direct mercury analysis (DMA). This method enabled recovery of >95 % of both mercury species from a single sampler, providing an operational solution for simultaneous speciation. An automated in situ cleaning system, the “miniwiper,” was designed to limit biofouling and particulate accumulation during long deployments, essential for detecting ultra-trace concentrations. Tested in productive Peruvian coastal waters and the particle-rich Nugu River (India), it reduced biofilm and particle deposition on DGT membranes. Under strong biofouling, it improved MMHg and HgII estimates, whereas uncleaned DGTs underestimated concentrations. In particle-rich river waters, miniwiper use yielded trace metal (Al, Mn, Fe, Co, Cd, rare earths) concentrations more representative of the dissolved fraction. Overall, this work demonstrates that DGTs are a promising tool for monitoring monomethylmercury and inorganic mercury at ultra-trace levels in marine waters. Further studies are needed to develop DGTs specifically optimized for HgII accumulation and to better assess miniwiper performance under diverse conditions. The protocols and devices developed open the way for integrating DGTs into autonomous observation networks and environmental monitoring programs recommended by the Minamata Convention.