The objectives of this study were to determine the trophic state of agricultural reservoirs within the four major watersheds and evaluate ecosystem health using a multi-metric fish modeling approach of the lentic ecosystem health assessment (LEHA) in South Korea. Fish survey for the LEHA model was sampled twice from 12 reservoirs (oligotrophic to hypereutrophic reservoirs) during 2008–2010 along with data analysis of water quality monitored by the Ministry of Environment, Korea. The fish sampling gears used in the field were fyke net, trammel net, casting net, and kick net. A regression analysis of N:P ratios showed a negative linear function with total phosphorous (TP) values but not with total nitrogen. The greater relationship to TP was due to low variability and high ambient concentrations of N relative to P. The plots of “TSI (CHL)–TSI (TP)” and “TSI (CHL)–TSI (SD)” suggest that other factors other than phosphorus limited algal biomass (CHL–TP < 0), and that non-algal particles dominated light attenuation (CHL–SD < 0). Nutrients input into the reservoirs during the Asian monsoon exceeded actual nutrient availability for phytoplankton growth. A regression model of the fish trophic components against log-transformed TP showed that the proportions of omnivore and insectivore fish had a positive linear function and negative relationship with TP, respectively. According to the multi-metric LEHA model for agricultural reservoirs, the model value averaged 26 ± 6.5, indicating a “Fair” condition. Overall, ecosystem health was directly influenced by organic matter pollution and high algal production. This new tool appeared particularly relevant to detect the contamination effects on fish communities in reservoirs. The new national biological monitoring methodology in lentic ecosystem would be used as a key tool for ecological restorations and species conservations in Korean lentic ecosystems.