The use of composite materials is of increasing importance over the past years. Especially unidirectional fibrous laminates are nowadays widely applied in industry. They provide mechanical advantages in terms of stiffness to weight ratios, strength and resistance against fatigue. These properties make them suitable for high-end applications as for example the aerospace industry. Topology optimization is a mathematical technique which has recently gained importance as well. As an optimization technique with a large design freedom, it is able to design complex structures with high performance beyond human abilities. Together with the latest improvements on manufacturing techniques, the application of topology optimized structures intensifies in various fields. This research focuses on topology optimization on unidirectional fibrous laminate structures. The problem of combined topology and fiber direction optimization is researched over the past years by a number of groups. The problem formulation where the fiber angles are directly used as design variables is highly non-convex and is likely destined to end up in a local optimum far from the global optimum. Two other alternatives are described in literature: a discrete and continuous problem formulation. In the discrete approach, called Discrete Material Optimization (DMO), a finite number of candidate materials per element represents the different fiber orientations and penalization is applied to end up with a clear distinction between the candidate materials. The discrete formulation has the drawback that the solution is limited to the predefined candidate materials and that the number of design variables easily becomes large. Furthermore, the global optimum could never be guaranteed due to the required penalization. The continuous approach uses lamination parameters as design variables and the optimization problem becomes convex. A shortest-distance approach is used to determine the closest realistic laminate configuration for the global optimal set of lamination parameters. Using this technique, continuous variable stiffness panels can be designed with a reasonable amount of design variables. However, the realistic laminate configuration to a set of lamination parameters is not known analytically for more complex problems. Therefore, the determination of a physically meaningful configuration may be a difficult task, and may go with a loss of performance. Given both the pro's and con's of the methods from literature, there seems to be a demand for a method that can provide detailed results (continuous variable stiffness), with a reasonable amount of design variables, which also directly provides a physically realistic laminate configuration. In this research a new method called the Adaptive Angle Set Method (AASM) is proposed. AASM solves a sequence of DMO-like subproblems for fiber angle optimization, but the associated design variables are not penalized. A separate set of density variables performs the topology optimization and the combined problem is solved simultaneously. Every subproblem in AASM is analogue to a non-penalized DMO problem with three candidate materials for every element, representing a set of three different fiber angfiles. In the initial subproblem, the angle set is equal for all elements and given by 60° 0° 60°, spanning the entire domain of 180° of possible fiber angles. This subproblem is solved to optimality and the subsolution is used to formulate the succeeding subproblem. Based on the subsolution of design variables, a combination of update functions estimates a new fiber angle for every element, which is defined as the middle angle of the element's new angle set. The two other angles are valued from this middle angle plus and minus a certain offset (range) and the new subproblem is again solved to optimality. However, the range between the three candidate materials is tightened with the formulation of every new subproblem, such that the sequence of problems converges to angle sets where the three candidate materials are close to each other. This can be as close as 1° difference in the final subproblem. At the final stage, penalization is applied to create a clear distinct solution between the candidate materials, but this only causes a minimal loss of performance due to the small range in the angle set. Using this approach, the number of design variables is constant for every subproblem, namely three fiber angle design variables and one density variable per element. In the final stage, a high angle resolution is obtained with a directly known laminate configuration. The way in which a new subproblem is formulated highly depends on the estimation of the new angle for every element. The determination of the optimal new angle using an optimization routine would be equal to solving the overall fiber angle problem, which can not be solved efficiently with a gradient based optimizer. Therefore, two heuristic update functions are introduced to estimate the new angle. The first update function makes a linear combination of the previous angle set with the corresponding optimal design vector. The second update function sets the new angle equal to the largest principal stress direction for that element. A number of test cases showed that a mixed application of both update functions yielded the best results. The final configuration was tested on a number of compliance minimization problems, which were kept planar and single loaded during this research. For small problems, the AASM results could be compared to brute force global optima of the underlying fiber angle integer problem. Results equal or close to the global optimum were obtained. For larger problems and multiple layer laminates, AASM provided promising results as well, which were obtained faster than a comparable DMO-formulation. The promising results obtained by AASM makes the method worthwhile for further investigation on larger and more complex problems, including other objective functions, bending elements and manufacturing constrained problems.