Data assimilation in space weather forecasting is necessary, but differs in important respects from terrestrial weather forecasting. These differences arise in part from the relative sparsity of data in space weather, from relatively direct driving of space weather phenomena, and from inherently unpredictable components arising from solar wind turbulence between the Sun and Earth. The differences between terrestrial and space weather assimilation requirements present opportunities for space weather forecasting to add new data assimilation techniques to those developed in terrestrial weather forecasting. This talk discusses these differences and presents a two-dimensional parameter space (data availability and sensitivity to boundary conditions) for comparing data assimilation requirements.