A new technique identifies parameterized quantum circuits that are more robust to noise. The work could improve the accuracy of quantum machine learning and quantum chemistry tasks, while using less computational resources in the circuit design process.