In this article I first briefly survey some enduring legacies of the Craik and Lockhart (1972) article on levels of processing (LOP) and address some common criticisms. In the next section I discuss whether memory can be regarded as ''pure processing'', the role of short-term memory in an LOP framework, measurement of ''depth'' in LOP, encoding-retrieval interactions, the concept of consolidation, and the reality of ''levels'' of processing. In the final section I offer some speculations on future directions, discussing the notion of levels of representation and a possible continuing role for LOP in memory research.