Extending the self-reference effect in memory to the level of social identity, previous research showed that processing information in reference to one's ingroup at encoding enhances memory for the information (i.e., the group-reference effect). Notably, recent work on the self-reference effect has shown that even simply co-presenting an item with self-relevant vs. other-relevant information (e.g., one's own or another person's name) at encoding can produce an "incidental" self-memory advantage in the absence of any task demand to evaluate the item's self-relevancy. In three experiments, the present study examined whether this incidental self-memory advantage extends to the level of social identity using newly created, minimal groups (Experiments 1 and 2) and pre-existing groups (Experiment 3; one's own or another study major). During encoding, participants judged the location of each target word in relation to a simultaneously presented cue (Ingroup-cue or Outgroup-cue in Experiments 1 and 3; Ingroup-cue, Outgroup-cue, or Neutral-cue in Experiment 2). Consistent across all experiments, a subsequent recognition test revealed a significant memory advantage for words that were presented with the Ingroup-cue. Crucially, this incidental ingroup-memory advantage was driven by ingroup-memory enhancement rather than outgroup-memory suppression relative to memory for words presented with the Neutral-cue (Experiment 2), and was positively correlated with self-reported levels of ingroup identification (i.e., self-investment to one's ingroup; Experiment 3). Taken together, the present findings provide novel evidence that mere incidental associations between one's ingroup and to-be-remembered items in a non-referential, non-evaluative encoding context can produce a memory advantage for the items.