Incident pain is a major problem for most patients with cancer pain. It is difficult to classify and poorly studied. The problem is that doses of opioids high enough to control episodic incident pain are too high when that pain is absent, leading to adverse effects. The commonest type of incident pain is pain on movement, leading to a quantitative increase in pain intensity. There is also a class of incident pain that is qualitatively different from any pain at rest, more akin to neuropathic pain. "Breakthrough" pain is the term used to describe pains that break through an existing analgesic regimen; these may be incident pains. There are no randomized controlled trials of incident pain management in cancer pain. The suggestions made for improving management are therefore based on such trials in other pain contexts. Studies in the cancer context will need to concentrate on characterizing the pain episodes.