In nuclear emergency management and long-term rehabilitation, dealing with uncertain information on the current situation, or predicted evolution of the situation, is an intrinsic problem for decision making. Uncertain information related to, for instance, incomplete information on the source term and the prevailing weather can result in dose assessments that differ dramatically from reality. Uncertainty is also an intrinsic part of model parameters. In the presence of uncertainty, ineffective decisions are often taken (e.g. too conservative or optimistic predictions, inadequately accounting of non-radiological risks), which may result in more overall harm than good due to secondary causalities as observed following the Chernobyl and Fukushima accidents. Therefore, the reduction of uncertainty, and how to deal with uncertain information, is essential to improve decision making for the protection of the affected population and to minimise disruption of normal living conditions.Decisions on early countermeasures (e.g. evacuation, sheltering, and provision of iodine tablets) are often taken based on deterministic calculations considering the best available source term information and the state of the art numerical weather prediction data as inputs. The source term uncertainty is large and quantification difficult. Uncertainty in meteorological forecasts is routinely quantified by "ensemble predictions". However, these predictions are not used in emergency management. One reason for this is the effort and time required to process 30 to 100 different weather sequences within atmospheric dispersion models.At present, decision making in the early phase is mainly based on modelling followed by a period where monitoring has made enough progress to provide a robust contamination and dose map. However, in between some monitoring data will be available; however, no operational tool exists to combine simulated dose estimations with monitoring data to obtain a more consolidated picture of the radiological situation.The key endpoint for all decision making is the dose and associated risk to affected population groups. However, in the case of the Fukushima accident, only order-of-magnitude dose bands were estimated from (local) contamination and model predictions, with no accounting for individual human behaviour patterns or location. Individual dose measurements could greatly reduce this uncertainty but are limited by capacity.Completely missing from emergency management is the assessments of risk in the early and the transition phases. However, this is important for decision making on the need for medical surveys and long-term support of the affected population.Operational decisions concerning land and food chain management rely on radioecological models that are mostly based on simple, but highly uncertain, transfer ratios to predict contamination in foodstuffs (though such data are few for some climatic areas, (e.g. Mediterranean) and foodstuff-radionuclide combinations). Process-based models might provide an approach to red...