Testing should stop when defined exit criteria (completion criteria) are met, ensuring objective quality goals are satisfied. Running out of time or tests is resource-based, not quality-based, and finding no faults may just mean tests are ineffective.