This study (i) presents a coupled vegetation-soil model adapted to perennial C4 seasonally flooded semi-arid grasslands; and (ii) applies the model to evaluate changes in the annual productivity of the grasslands of the Tana River Delta, Kenya, under changing flooding conditions. Main plant growth processes are modelled within coupled plant carbon balance and soil water budget modules: photosynthesis, allocation of photosynthates, respiration, translocation of root phytomass to aerial phytomass, senescence and litter production. Aerial phytomass can also be subtracted from the system, to simulate grazing or cutting. New features concern the inclusion of effects of floods on energy conversion efficiency, photosynthate allocation, senescence and litter production. The vegetation model, composed of four phytomass compartments (leaves, stems, roots, aerial dead matter), simulates three main growth phases related to flooding: floods, a post-flood phase and a non-flooded phase. It was designed to be used with limited climatic data. Data collected during a 14-month experiment (2010–2012) in the Tana River Delta, Kenya, in which different irrigation and cutting treatments and flood events were recorded, were used for calibration and validation purposes. Fourteen parameters, selected through a sensitivity analysis, were calibrated on half of these treatments. Uncertainty in parameter estimation was expressed through a stochastic ensemble of simulations. The remaining independent data were used for model validation. Overall, the model predictions are in good agreement with the experimental data. This model can be used to assess the impact of rain variability, grazing or flooding patterns on the annual primary productivity of Sub-Saharan floodplain grasslands composed mainly of Echinochloa stagnina (Retz) P. Beauv. In particular, simulations for the Tana River Delta suggest that past changes in the hydrological regime of the river, as well as future changes due to the construc