Principle component regression (PCR), artificial neural network (ANN) and their combination used as data driven models were proposed in this study to forecast the phytoplankton dynamics in Macau Main Storage Reservoir (MSR) that is experiencing algal bloom in recent years. The models were based on comprehensive 8 years (2001-2008) monthly water quality data for training, and the most recent 3 years (2009-2011) monthly data for testing. Twenty three water quality variables including physical, chemical and biological parameters were involved, and comparisons were made to select the best model that are appropriate to MSR. Our simulation results showed that, in contrast to PCR, the present work using ANN has better accuracy performance and generalization performance (R2 = 0.758, 0.760) (the 1st number denotes for training and the 2nd number denotes for testing) over the total random 50 runs. Using principle component analysis as preprocessing (PCA-ANN) did not show better performance than using ANN alone, implying that eliminating the uncorrelated variables do not increase the prediction powers. Moreover, contradictory to the previous studies showing that the hybrid model can handle both linear and nonlinear components of the problems well, the PCR-ANN in this study obtain no better improvement. In spite of still some rooms in the model, the present work using ANN has been proved to be a feasible approach to forecast phytoplankton population in MSR.