We studied the eutrophication history of a tropical shallow reservoir in the São Paulo metropolitan region, southeast Brazil. We analyzed grain size, geochemistry, diatom assemblages, and land-use records in a sediment core from the reservoir to infer its trophic state history during the last ~110 years (1894–2005). Eighty diatom species were observed in the core and shifts in the relative abundances of planktonic and benthic taxa indicate major limnological changes associated with complex interactions between hydrologic factors and eutrophication. Discostella stelligera was associated with deforestation and water physical changes whereas Aulacoseira granulata, a species abundant throughout the core, was mostly associated with high flux conditions and erosion events, regardless of trophic state. Eutrophication was triggered by construction of the city zoo (1958) and installation of the São Paulo State Department of Agriculture (1975) within the Garças watershed, and increasing loads of untreated sewage from these institutions. The data suggest that deterioration in water quality began after ~1975 and markedly accelerated after ~1990. The reservoir has been hypereutrophic since 1999. Steady increases in geochemical proxies for trophic state, along with a decrease in C/N ratios, indicated higher nutrient concentrations and the prevalence of autochthonous production towards the core top. Appearance of Achnanthidium catenatum ~1993 highlighted the onset of a marked eutrophication phase. The subsequent dominance of Planothidium rostratum and Cyclotella meneghiniana suggested a sharp shift to a hypereutrophic state since 1999. Land-use history proved valuable for validating the chronology and interpreting anthropogenic impacts. Multi-proxy analysis of the sediment record provided an effective tool for tracking ecological shifts in the reservoir ecosystem. This study provides the first reconstruction of lake eutrophication history in Brazil and highlights the importance of hydrological/physical changes as drivers of diatom assemblage shifts in reservoirs, which may confound trophic state inferences based on shifts in the planktonic/benthic diatom ratio.