Lake and reservoir managers are in need of simple tools for predicting harmful algal blooms. But the process-based models described in the literature are often highly complex and require large data sets. Beyond a certain degree, adding processes was shown to reduce model predictive capabilities. In this work, we assess the performance of a simple biological model to describe the succession of cyanobacterial blooms in a reservoir. Our study site is Karaoun Reservoir, Lebanon, a eutrophic reservoir where cyanobacteria Aphanizomenon ovalisporum and Microcystis aeruginosa alternatively bloom. We used the following configuration of the one-dimensional hydrodynamic-ecological model Dyresm-Caedym: both cyanobacteria were simulated, with constant buoyancy for Aphanizomenon ovalisporum, vertical migration dependent on light and internal nitrogen for Microcystis aeruginosa and growth limited by light and temperature for both species. The model was calibrated in summer and autumn 2012 and validated in spring and summer 2013. It showed a good performance for the water level (RMSE 1 m, annual variation 25 m), water temperature profiles (RMSE 1°C, range 13-28°C) and cyanobacteria biomass (RMSE 48 µg L-1 equivalent chlorophyll a, range 0-206 µg L-1). This shows that simple model configurations can be sufficient when few major processes can be identified. This approach could be transposed on other eutrophic lakes and reservoirs to describe the competition between dominant phytoplankton species, contribute to early warning systems or be used to predict the impact of climate change and management scenarios.