Repetitive construction projects allow for planning to be done systematically, but performance is hard to predict due to inherent uncertainties. The goal of this research is to evaluate planning choices in repetitive construction projects. In order to evaluate the planning of these projects to quantify the risk of overruns, an adaptable digitized model is needed which can run realistic probabilistic simulations. The model was supplemented with realistic effects such as probabilistic distributions and learning curves to represent the inherent uncertainties in the progression of the project. The model is able to estimate the total duration of a project, the time-dependent costs and resource utilization percentage. Even more so, the model was used to give an estimation of the risk of overruns as a probability. The model has been used to evaluate decisions that can be taken when planning a project. To test how these planning decisions affect the project outcomes, they were tested in a test case set-up. The influence of one planning decision was increased and the outcomes measured. These planning decisions consist of 1) the use of multiple resources crews per process, 2) the use of batching, 3) using time buffers and 4) multiskilled crews. For decreasing the project duration, extra crews can be added. Time-dependent costs are decreased by adding buffers or working with larger batches. Improving the utilization percentage is done by using multiskilled crews and adding buffers. The results are relative and can be interpreted as rules of thumb, but are indicative of the power of planning decisions to steer project performance without altering the scope or the construction method. To show that the model indeed corresponds to these rules of thumb, they were applied in two projects, yielding promising results, both for total duration and time-dependent costs. Using the understanding of the effects of planning decisions, repetitive construction projects can be improved to increase productivity and decrease time-dependent costs significantly.