The Matlab code accompanies the work "Confidence Regions for Simulations with Learned Probabilistic Models" by Armin Lederer, Qing Hao and Sandra Hirche accepted to the American Control Conference 2020. The code is as is and is not maintained. Execution of the code requires MATLAB R2017a, the Optimization Toolbox, the Statistics and Machine Learning Toolbox, and the Parallel Computing Toolbox.