System Dynamics and Multi Criteria Decision Analysis are two decision aid methods that aim at helping decision makers to better understand the world through formal models and computations. System Dynamics with the use of feedback loops and time delays helps to understand the behavior of a system and how the various elements affect this behavior. Moreover, it can be used for testing policies in a safe, consequence-free environment. However, one of the disadvantages of SD is that it is often difficult for the decision maker to determine which policy is the most appropriate. Multi Criteria Decision Analysis (MCDA) is a method that provides decision makers with tools to establish objectives, clarify criteria that represent their preferences, attach levels of importance to those criteria and finally assess policies in a more structured way. However, there are a lot of MCDA methods making it difficult to choose the most appropriate one. Moreover, it is often difficult to establish the values for the criteria that will be used to evaluate the alternatives/policies. Furthermore, uncertainty is inherent in decision making. Several sources of uncertainty could be identified such as the difference between the system “as is” and “as it is perceived to be”, the errors in the data and the uncertainties that are inherent in every method (parameters, functions etc.). As a result, uncertainty cannot be avoided in the decision making process. The aim of decision aid though, is to help decision makers identify robust alternatives in the face of uncertainty. Robustness is a term with many definitions in the decision making literature. In this master thesis, robustness is a property of an alternative that makes it valid for all or most versions of the process, where most versions means the combinations of parameters, choices etc. One way to mitigate several sources of uncertainty and disadvantages of SD and MCDA is the combination of the two methods. Their nature seems complementary: SD is used for understanding a system and testing policies and MCDA is used for policy comparison and selection. However, not many efforts have been made towards that direction. Moreover, the efforts that were made did not address issues like taking into account different perceptions of the system in question, evaluation of the alternatives over time and how to distinguish robust alternatives in spite of the presence of uncertainty. One method to address issues of deep uncertainty, is Exploratory Modeling and Analysis (EMA). EMA adopts modeling and simulation processes that are not used merely for predictive modeling, but mostly for better understanding of a system and its uncertainties, under different levels of detail and perception. The purpose of this master thesis, is the development of a methodology (along with the development of a computer program) that will combine SD and multiple MCDA methods with the purpose of testing and assessing policies under different perceptions, levels of detail and to reduce the uncertainties deriving from any single method when used alone. For the development of the program several choices had to be made. First, the output of the SD models will serve as the input for the MCDA process. Each policy will be simulated and its results will be used as the data for the MCDA. Second, several decision makers and their preferences could be represented in the program. Each stakeholder can provide different criteria, different values, weights and (utility and preference) functions. Furthermore, the policy assessment will occur for different points in time, since the output of SD shows behavior over time. Moreover, several MCDA methods were studied in the literature and evaluated under the criteria of familiarity with the method, covering of the classifications proposed by Figueira et al. (2005) and finally how much they are used in the literature. The chosen MCDA methods are: performance targets, Multi Attribute Utility Theory and PROMETHEE II. For these methods, each stakeholder/decision maker can provide a range of weights, different utility/preference functions with a range of values for each (function) parameter. Finally, to search for robustness among the different rankings that will be generated the following process was used: for each ranking, each alternative was tested if it falls within 75% of the alternative with the maximum score in the particular ranking. If it did, the alternative became member of a group. Otherwise, the alternative became member of a second group. In the end, the number of appearances in each group is calculated and the alternatives with the most appearances in the first group are considered the most robust. To test and illustrate the program, a model by Tsaples et al. (2013) was used. The SD model deals with the implementation of land value taxation and its consequences in the development of a city. Different taxation regimes were simulated. The different taxation regimes were then tested in the program. The results demonstrated that the combination of SD with MCDA could help decision makers by identifying not only the alternatives that are consistently robust (in a “good” or a “bad” way), but also which alternatives could be potential points of friction among the stakeholders. However, the program did not come without disadvantages. Mainly, the large number of the generated rankings could mean an overflow of information that could result in performance downgrade. Moreover, the program that was developed for this master thesis can be considered a pilot program. Thus, the choices that were (and were not) made might not be ideal and several additions and corrections are necessary. In addition, the program itself provided some insights on the decision making process and the tools that are used to facilitate it. First, the decision making process could be broken down in pieces and different methods could be deployed to analyze every part. Second, even within each method, an exploratory approach with the use of multiple perceptions, points of view and parameter values could mitigate some of the disadvantages that are inherent in each method. The inclusion of multiple stakeholders ensures that the system of interest is studied under different perceptions. Thus, the effort should be focused not on making the decision makers agreeing but on finding solutions that seem appropriate under all those perceptions. The time problem in MCDA lingers as problem in the search for robustness. The static nature of the method means that so far, alternatives were assessed on the specific point in time, although in real life situations, policies are (or should be) assessed over a period of time. Consequently, there is a need for further study on what time actually means for the decision aid tools and how it could help improve those tools. In addition, by assessing policies over a period of time, robust alternatives could be considered not only those that address the specific issues for which they were designed, but also those that change successfully over time to do so. Furthermore, the notion of robustness needs further clarification and a documentation of its properties and levels could be the first step towards the development of methods that will more clearly identify policies that are valid in almost all versions of the computations. Finally, the exploratory approach on modelling needs further studying. The rise of computational power has made it more feasible to use it than a few years ago. Although the exploratory approach does not come without its disadvantages, its use could offer great benefits in the decision making process. This thesis is a demonstration of that potential.