Soybean (Glycine max., L. Merr) seed yield was simulated by combining the concept of hydraulic model of pod set, seed number determination driven by assimilate partitioning, and phenology driven mean individual seed growth. Plant dry matter production was simulated using modified approaches of (1) sink and source driven leaf growth, (2) C3 leaf photosynthesis, (3) multi-leaf-layer canopy photosynthesis integration, and (4) gross and maintenance respiration. Phenology was simulated as function of temperature and daylength. The model accurately simulated the accumulation of total above ground biomass and seed yield in two cultivars with contrasting stem termination types (semi-determinate and indeterminate) and in two years (2004 and 2005). The effects of population density and cultivar on above ground growth and seed yield are accounted for. Under optimum conditions allowing dry matter accumulation up to 15 t/ha, the model agreed with the measured data, Harvest index was negatively correlated with seed yield.