Performance testing mistakes include running tests for too short a duration (insufficient to reveal steady-state behavior or memory leaks) and using insufficiently populated test data (fails to represent real-world volume and variety). Tests that are too long or have adequate data are not mistakes - they're good practices.