Epistemologists often remark that knowledge precludes luck. A true belief based on a guess or hunch is not knowledge because it seems merely fortuitous, too much of an accident, and, well, lucky that one happened to get things right. Of course, true beliefs based on guesses and hunches are not justified. However, Gettier cases have persuasively shown that even justified true beliefs can admit knowledge-precluding kinds of luck. So in what sense are justified true beliefs that don’t amount knowledge lucky or ‘true only by chance’? I will address three different approaches to epistemic luck that have received significant attention in the recent literature: reliability approaches to epistemic luck, credit approaches to epistemic luck, and responsibilist or control approaches to epistemic luck. I will then turn my attention to the implications that epistemic luck has for other epistemic states. I will focus particularly on state of understanding and examine whether, and to what extent, understanding is vulnerable to epistemic luck.