How might use cases and use case testing be an effective black-box testing technique?

How might use cases and use case testing be an effective black-box testing technique?

Why or why not, and in what situations?