The objective of this study was to construct a spatial simulation model of the vegetation dynamics in semi-permanent prairie wetlands. A hydrologic submodel estimated water levels based on precipitation, runoff, and potential evapotranspiration. A vegetation submodel calculated the amount and distribution of emergent cover and open water using a geographic information system. The response of vegetation to water-level changes was based on seed bank composition, seedling recruitment and establishment, and plant survivorship. The model was developed and tested using data from the Cottonwood Lake study site in North Dakota. Data from semi-permanent wetland P1 were used to calibrate the model. Data from a second wetland, P4, were used to evaluate model performance. Simulation results were compared with actual water data from 1797 through 1989. Test results showed that differences between calculated and observed water levels were within 10 cm 75% of the time. Open water over the past decade ranged from 0 to 7% in wetland P4 and from 0 to 8% in submodel simulations. Several model parameters including evapotranspiration and timing of seedling germination could be improved with more complex techniques or relatively minor adjustments. Despite these differences the model adequately represented vegetation dynamics of prairie wetlands and can be used to examine wetland response to natural or human-induced climate change.