Within the EURANOS project data assimilation (DA) approaches have been successfully applied in two areas to improve the predictive power of simulation models used in the RODOS and ARGOS decision support systems. For the areas of atmospheric dispersion modelling and of modelling the fate of radio-nuclides in urban areas the results of demonstration exercises are presented here. With the data assimilation module of the RIMPUFF dispersion code, predictions of the gamma dose rate are corrected with simulated readings of fixed detector stations. Using the DA capabilities of the IAMM package for mapping the radioactive contamination in inhabited areas, predictions of a large scale deposition model have been combined with hypothetical measurements on a local scale. In both examples the accuracy of the model predictions has been improved and the uncertainties have been reduced.