According to a familiar view in philosophy of mind, mental states or properties are realized by physical states or properties but are not identical to them. This view is often called realization physicalism. But what is realization? I argue that recent approaches to realization, represented by Carl Gillett's 'dimensioned' view, fail to acknowledge some textbook cases of realization. I also argue Gillett's account in particular admits realization relations that should not count if realization physicalism is to be distinguished from its competitors in the usual ways. I offer my own account of realization, and argue that it is superior not only in passing the above tests but also in its utility for answering questions about multiple realizability.