Optimizing business processes can potentially save costs and improve efficiency and quality. Different optimization methods have been proposed for the control flow aspect, but for the resource aspect, there is still much potential for improvement. Concerning optimization of resources, questions arise, such as the optimal number of employees, that is needed next week, given a set of constraints, for example the customer should not have to wait longer than two weeks. Various analysis and optimization methods are developed and evaluated in this dissertation to answer this type of questions. In particular, two analysis methods are developed, one based on simulation and one based on queueing theory. In addition to that, a method is developed that can be used to efficiently find the optimal number of resources that should be used in a business process, given certain constraints. Simulation is a common analysis method for business processes, but current business process simulators lack several properties to ensure a reliable result. Issues like lack of replications or the absence of the support for probability distributions can lead to incorrect results. Furthermore, common resource properties are often not supported, such as separation of duties, case handling or allocation strategies. Using the new simulator, the quantitative effects of the lack of support for the resource properties is evaluated. Significant differences indicate that the simulation outcome would not represent the actual process performance if these resource properties are not supported. Using the newly developed simulator, more accurate results can be obtained for the performance of a business process, given that the resource properties are supported in this simulator. Optimization of resources in a business process can be very complex. Since there are several types of resources, different performance of resources and multiple periods to optimize, the optimization solution space can be very large. To explore the search space for the optimal value, the analysis of the process given the selected resource setup is executed many times. This requires the analysis method to also be fast. While the developed simulation tool is very accurate, it lacks speed. A single simulation run can easily take minutes or longer, which – in a large enough search space – can lead to very long optimization calculations. A queueing analysis method is developed to solve this problem, which also supports constructs that existing queueing analysis methods do not support, such as parallel paths and roles. The evaluation shows that the new queueing analysis technique performs better in all cases than the state-of-the-art, where for the parallel construct, an improvement of 50 percent in the error compared to the simulation model is achieved. The advantage of queueing model analysis over simulation is that the computation time is reduced from seconds or minutes to only milliseconds and hence makes it a viable option to use in optimization problems with a large search space. The two developed analysis methods can be used to analyze the performance of the business process given a resource configuration. To find the optimal solution, the search space is traversed for the resource configuration which is within the constraints and has a minimal number of resources. To reduce the number of steps taken in a naive search, which search the complete search space, three versions of a search strategy are developed. These search strategies are compared against each other and against a naive search strategy. Using a guided search based on the resource utilization rate, the search space is traversed more efficiently. An even more efficient way of searching using binary search is also developed, but this might lead to sub-optimal solutions. Therefore a third variant is developed combining binary search with a local search near the final solution, to increase the probability that the optimal solution is reached. All three search strategies are evaluated on their performance on different search spaces. The performance of the methods is independent of the process structure, but the search space’s size influences how quickly a strategy reaches the optimal resource configuration. Using the search strategies the optimal resource configuration for a business process can be determined in an efficient manner. The number of steps to compute the optimal resource configuration for a real-life size process is reduced from 125.000 steps to only 22 steps. These search strategies provide a basis for further expansion of the optimization of business processes with, for example, extending to resources involved in multiple processes or multi-period optimization.