We begin with a theoretical overview of the concepts of recollection and familiarity, focusing, in the spirit of this special issue, on the important contributions made by Andrew Mayes. In particular, we discuss the issue of when the generation of semantically-related information in response to a retrieval cue might be experienced as recollection rather than familiarity. We then report a series of experiments in which two different types of masked prime, presented immediately prior to the test cue in a recognition memory paradigm, produced opposite effects on Remember vs. Know judgments. More specifically, primes that were conceptually related to the test item increased the incidence of Remember judgments, though only when intermixed with repetition primes (which increased the incidence of Know judgments instead, as in prior studies). One possible explanation--that the fluency of retrieval of item-context associations can be experienced as recollection, even when the source of that fluency is unknown--is counter to conventional views of recollection and familiarity, though it was anticipated by Andrew in his writings nearly two decades ago. (Contains 4 tables, 8 notes, and 7 figures.)