This document introduces MUBench, a benchmark for evaluating API misuse detectors. MUBench contains 89 real-world examples of API misuse collected from open source projects on GitHub and SourceForge, as well as from a survey. Each misuse example includes a description, the types involved, and metadata linking to the project and fixing commit. The benchmark is hosted on GitHub and aims to help researchers test and compare different approaches for detecting API misuses.