Serial memory refers to the ability to recall a novel sequence of items or events in the correct order. In the laboratory, the dominant tool used to assess this mental faculty is the immediate serial recall (hereafter, ‘serial recall’) task in which participants are given a sequence of typically verbal, visual, or spatial items that they must subsequently recall in their original presentation order. Serial recall is a deceptively simple task—the apparent ease with which people accomplish it masks the wealth and complexity of findings this task has generated, and the computational theories that have been developed to account for them. In this chapter, I review benchmark findings of serial recall that have been observed across the verbal, visual, and spatial short-term memory domains, and I interpret them with reference to the core mechanisms embodied in contemporary computational theories of serial recall. This analysis identifies four mechanisms that are common to the three content domains—namely, position marking, a primacy gradient, competitive queuing, and response suppression. Additionally, evidence suggests that in verbal serial recall both the encoding and retrieval of items is sensitive to item similarity—similarity-sensitive encoding and retrieval—and that item retrieval is accompanied by output interference. By contrast, in visual and spatial serial recall there is evidence for similarity-sensitive retrieval, but the relevant empirical observations that evince similarity-sensitive encoding and output interference are yet to be studied in the visual and spatial domains. I conclude by outlining some challenges for future research.