Satellite remote sensing data are a key source for the systematic monitoring of inland waters. The current availability of medium and high spatial resolution sensors brings new opportunities for mapping small water bodies. However, inland waters may be subjected to the adjacency effects which affect the radiance leaving the water surface due to the photons scattered from the surrounding land targets. These effects impair the accuracy of water constituents' retrieval because the scattered radiation from neighboring land targets gets into the sensor's path through the atmosphere and is added to the sensor’s signal. This complex phenomenon affects the surface reflectance retrieval, and its correction is a requirement for the quantitative application of satellite-imagery on inland waters. The objective of this research was to evaluate occurrence and correction of the adjacency effect on inland waters, using a medium spatial resolution sensor (MSI/Sentinel-2 A and B). The study area included five small lakes surrounded by dense forest cover located in the Amazon region and one large urban water reservoir in the São Paulo State, Brazil. In this research, three main analyses were conducted: (i) the application of a convergence method to estimate the aerosol loading at 550 nm (AOD550) using in-situ water reflectance measurements as reference in the inversion of the radiative transfer equation; (ii) the assessment of the physical method performance based on Atmospheric Point Spread Function (APSF) for adjacency correction on inland waters. Three approaches to recover the size of horizontal range of the adjacency effect (HAdj) were assessed: Fixed window, SIMilarity Environment Correction (SIMEC), and Adaptative Window by Proportion applied to Inland Water (AWP-Inland Waters). AWP-Inland Water is a preliminary algorithm developed in this research based on the proportion of non-water targets within the window; and (iii) the assessment of the adjacency effect sensitivity to environmental factors using theoretical simulations. The accuracy assessment of the adjacency correction using the HAdj approaches and Atmospheric Correction (AC) was performed with in-situcollected samples along the selected water bodies (N=46). With optimal AOD550values, the AC presented a good agreement, especially at the visible wavelengths, with the validation data for all investigated water optical types (MAPE: eutrophic ~56%, bright ~80%, and dark ~288% waters) when compared to common sources of aerosol loading extraction, such as MODIS-products (MAPE: eutrophic ~73%, bright ~105%, and dark ~402% waters). However, the inversion model does not work well when its assumptions are not satisfied. By examining the retrieval of the atmospherically corrected water reflectance values, both methods (MODIS and inversion model) showed uncertainties in obtaining accurate reflectance values in the near-infrared wavelengths due to adjacency effects. Regarding the adjacency correction, the estimated water reflectance was associated with smaller errors from the AWP-Inland Water method, considering only dark waters (MAPE: ~53%). The adjacency correction performance in eutrophic and bright waters was similar using all HAdj methods. SIMEC and Fixedwindow presented a strong trend to produce invalid results (i.e., negative water reflectance values) at the near-infrared wavelengths due to the overestimation of HAdj size when applied to small dark water bodies under very high adjacency effect. Significant errors produced by the adjacency correction from SIMEC and Fixed window invalided their application in dark waters and small water bodies. Simulated results demonstrated that several factors could influence the adjacency effect magnitude, such as the shape and size of water bodies, aerosol properties (e.g., aerosol loading and aerosol model), proportion of non-water targets within the HAdj, land cover around the water body, and water composition variability (e.g., events of algal-blooms). In general, the adjacency effect is maximized for small water bodies, higher aerosol loadings (more than 0.1), and dark waters (water reflectance less than 4%). For example, in this critical arrangement, the adjacency contribution (~53%) at the Top of Atmosphere (TOA)was up to ~5 times larger than the water contribution (~11%) at 740 - 842 nm wavelengths. This research contributes for further understanding of adjacency effects in medium spatial resolution imagery on inland waters, using a physical based approach, including the uncertainties in the HAdj determination, which still remains a challenge for next studies.