Thermal testing - chamber vs real-world temperature cycling?

Chamber temperature cycling is repeatable but real-world profiles are messier. Did chamber testing ever miss a failure that showed up in the field, or vice versa?