I have had mostly bad experiences with PowerMock. As Todd says, it is usually indicative of something that needs to be refactored or fixed; or is simply mis-used.
Tests that use too much PowerMock are extremely brittle, often failing in test code B after after changes in seemingly unrelated code A in ways that require deep knowledge of B. It does some funky stuff in classloader land that results in painful to debug issues. For example, I have seen static code initialized twice, leading to a test that hangs forever.
We have started to use phrases like "friends don't let friends abuse PowerMock". Fixing a broken PowerMock test takes much longer than fixing an ordinary test.
It is a useful tool to have around, but needs to be used judiciously. Every time it is used "because I must mock XYZ" ask, "why does XYZ need to be mocked? does it need changes to design to be testable?"