A relationship has been hypothesized between metamemory (the self-knowledge of memory processes that an individual can verbalize) and actual memory performance. To explore this relationship, a study was conducted, in partial replication of earlier studies, using the strategy choice paradigm for paired-associate learning (PAL) on a metamemory questionnaire. Metamemory subtests were administered to second graders (N=66) at the beginning of each session. Three PAL trials followed with each list comprising 10 known pairs presented over one study-test trial. The first two trials familiarized subjects with an effective sentence elaboration strategy and an ineffective rote repetition strategy. In the third trial, Choice, subjects freely selected a learning strategy. The results revealed that Elaborators (N=44) on the Choice-trial learned three times more material than did Repeaters (N=22). The metamemorial knowledge of Elaborators concerning the acquisition process was significantly greater than that of Repeaters, while scores on metamemory subtests related to retrieval processes were comparable for Elaborators and Repeaters. These results suggest that a fair test of the metamemory-memory connection requires an explicit definition of the metamemory construct in terms of its functional relation to either acquisition or retrieval skills, and that effective tests of the metamemory-memory connection require the maintenance or transfer of elaborative strategies rather than the spontaneous use of familiar learning strategies. (Author/NB)