Dynamics of phytoplankton (chlorophyll a, an index of biomass) and water quality in the Youngsan River estuary, S. Korea were analyzed to identify main mechanisms controlling the dynamics using artificial neural network (ANN) modeling. Ten years (2008-2018) of the long-term data were used including size-fractionated (micro-, nano- and pico-sized) chlorophyll a, freshwater discharge, water temperature, salinity, surface-bottom difference of temperature and salinity, water transparency, PAR, solar radiation, nutrient concentrations. The network consists of three layers: input layer of 18 neurons, hidden layer of 24 neurons, output layer of 4 neurons which are the output variables (whole, micro-, nano- and pico-sized chlorophyll a). The network was trained using an error backpropagation training algorithm and validated (R²≥0.997) by comparing the estimates with the observed results. Input data were transformed by Weibull distribution probability function and rescaled into a [0, 1] interval before training the neural networks to identify main mechanisms controlling the size-fractionated phytoplankton. The results of modeling analysis suggested that water temperature and salinity are important in determining the size structure of phytoplankton in the Youngsan River estuary, S. Korea.