The topic of working memory (WM) has increased dramatically in citation counts since the early years, not all of course related to or supportive of my own work, but a recent attempt to review it ended with more than 50 pages of references. What follows is a partial, as opposed to impartial, account of the origins of the concept of multi-component working memory (M-WM) and of the author's own views on its subsequent development. He then focuses on the phonological loop on the grounds that it seemed the most tractable system to investigate, given the very extensive earlier research on verbal STM. He then viewed the phonological loop as a relatively modular system comprising a brief store together with a means of maintaining information by vocal or subvocal rehearsal. In the 1960s, a number of studies attempted to decide whether forgetting in the STM system was based on trace decay or interference.