Urbanization is a global phenomenon observed at unprecedented pace and scale. It is a threat to many ecosystem services and yet it also offers the possibility to shape a sustainable future. The United Nations have defined 11 goals that are intended to make cities and communities sustainable. Reaching these goals is a wicked challenge for decision-makers because the goals are often conflicting and there are many uncertainties and complexities that need to be considered when addressing them. In order to help decision makers to deal with a wicked challenge, to anticipate and understand the consequences of their actions or even to help them develop a vision for the future, it is often necessary to employ simulation models. A wide variety of modelling and simulation approaches dealing with urban development is available. However, many of them are not very suitable to deal with the wickedness of the problem of reaching the sustainable urban development goals. In particular, they may struggle to deal with the many conflicting objectives involved. Using a multi-objective optimization approach specifically offers the possibility to account for conflicting objectives. It offers potential solutions to problems that are too complex for humans to solve, and it can be used to derive optimal solutions comprising a vision for a sustainable future. Thus, multi-objective optimization has some interesting properties that could be exploited and used to complement the existing set of modeling approaches used to support decision-making related to sustainable urban development. However, so far the possible applications of multi-objective optimization have rarely been demonstrated to support decision-makers dealing with the wicked challenge of improving the sustainability of urban development. The aim of this thesis is not only to offer support to decision-makers dealing with the sustainability of urban development, but to demonstrate how multi-objective optimization can be used to support decision-makers for a larger scope of problems related to land-use change and management. The latter is supposed to promote the use of multi-objective optimization in land-use system science. A further step in promoting multi-objective optimization in land-use system science is to present effective algorithms to solve spatially explicit multi-objective optimization problems, which is a prerequisite for applying multi-objective optimization. I exemplify the use of multi-objective optimization by addressing two goals that are related to sustainable urban development. These two goals (i.e. objectives) are to maximize compact urban development and to minimize the loss of fertile agricultural soils in Swiss municipalities. In chapter 2, the use of multi-objective optimization will be promoted by showing that there are efficient ways of solving spatially explicit optimization problems. As a strategy for solving the optimization problem I use a so-called genetic algorithm, which is a popular approach when dealing with two objectives. Promoting a wider use of multi-objective optimization approaches, by applying them to answer new research questions and showing possible new applications is achieved in chapters 3, 4 and 5 of this thesis. In more detail, in chapter 2 I demonstrate how genetic algorithms (GAs) can be modified in order to solve complex multi-objective optimization problems involving spatial relationships. I used and adapted, the so-called NSGA-II (Non-dominated Sorting Genetic Algorithm – II), for solving the multi-objective problem of minimizing the loss of fertile agricultural soils (i.e., agricultural productivity) due to urban growth and at the same time to maximize compact (i.e. contiguous) urban development. I compared existing modifications of GAs from literature and modifications developed by myself. In order to account for the spatial relationships, I found that it was crucial to include knowledge about how compact urban patterns evolve. In contrast, objectives such as agricultural productivity that do not involve any spatial dependencies (i.e. no neighborhood-relationships are involved) do not need to be dealt with in a special way. This knowledge provides a guideline for future researchers on how to solve multi-objective optimization problems involving spatial relationships. In chapter 3, I show that it is possible to efficiently reduce the loss of agricultural productivity by steering the pattern of urban growth. In a first analysis, I reveal that fertile soils are often found in the vicinity of existing urban areas. This leads to a potential trade-off, as compact (i.e., contiguous) urban development can thus lead to a high loss of fertile soils. In order to protect as much agricultural productivity (i.e., fertile soils) as possible, decision makers can either resign from having compact urban development or can strive for solutions (i.e. urban patterns) that were obtained in an optimization process. Although using optimization is a difficult and costly approach, my results show that this approach can be used in an efficient manner. Firstly, by comparing simulations of Business As Usual (BAU) urban expansion and urban patterns obtained by multi-objective optimization, I was able to show that there exist some regions for which there is a large difference between BAU and optimal solutions (methodical details on how the BAU urban expansion was modelled can be found in Appendix 1.). In order to aid decision makers, I derived a simple rule stating that in regions where high urban growth is anticipated (i.e. many agricultural areas will be converted into urban) the difference between BAU and optimal solutions is large and policy-makers should focus their efforts on steering the pattern of urban development in these regions. Secondly, there are areas of agricultural land that can be converted into urban without losing compactness or the most fertile soils. This knowledge could help policy-makers and planners prioritize areas that can safely be converted to urban without destroying more fertile soils than necessary or without losing more compactness than necessary. In chapter 4, I use multi-objective optimization in order to simulate zoning decisions of urban planners and I exemplify how planning paradigms (in the form of constraints) can be an obstacle in reaching optimal solutions. Although there are many objectives that urban planners need to account for, I am assuming that they focus mainly on reducing the loss of fertile soils and promoting compact urban growth. The planning paradigms that I consider is that zoning is carried out at a local level (i.e. at municipality level) and that a predefined amount of urban zones needs to be created in each municipality. My results show that planning at a local level can be a constraint in reaching more optimal solutions and that cooperation at higher levels is required for an improved protection of agricultural productivity. While the latter is not surprising, I further show that using multi-objective optimization can elucidate whether and how municipalities should cooperate. The two ways in which municipalities can cooperate are (1) that they adapt their preferences in collaboration with other municipalities and (2) that they make agreements on the amounts of urban growth that is permitted in each municipality. The first option for collaboration can be useful if, e.g., two municipalities adapt their preferences, because in one of them a slightly higher loss of compactness can lead to a large gain in agricultural productivity, while in the other one a small increase in the loss of agricultural productivity can lead to a large gain in compactness The second option for collaboration is more effective than the first option, i.e., leading to a stronger reduction of loss of agricultural productivity. However, the second option may require a stronger institutional framework than the first one and may thus also be related to disadvantages. In chapter 5, I use multi-objective optimization to simulate consecutive planning periods in order to come up with recommendations on the length of planning horizons. Again I am relying on the example of allocating urban zones in such a way that compactness of the derived urban patterns is maximal and the loss of agricultural productivity is minimal. In a toy experiment, I prove that due to the non-linear combinatorial nature of the problem of optimally allocating urban zones, short planning horizons may lead to non-optimal solutions. However, in a real-world situation I show that a short planning horizon does not necessarily lead to non-optimal patterns. While these results are interesting, the approach described in chapter 5 is especially valuable for demonstrating a methodology that could be enhanced in order to study in more detail the adequate length of planning horizons depending on the objectives involved and the characteristics of the planning perimeter. In summary, this thesis shows that multi-objective optimization can complement current urban- and land-use modelling approaches by answering novel questions or questions that remained unanswered so far. The methodological advances presented in chapter 2, 3, 4 and 5 will help future researches to understand whether multi-objective optimization may be an appropriate approach to address their research questions. Pursuing the same purpose, i.e., facilitating researchers with a better understanding on the possibilities and limitations of multi-objective optimization, chapter 6 discusses a potential taxonomy of optimization approaches for urban and land-use modelling. The same chapter concludes with possible future research directions that build upon results presented in this thesis and with some recommendations concerning sustainable urban development.