The role of meaning in the learning of verbally presented paired associates (PAs) is described by means of an experiment with a replication. A printed PA list containing six different types of items was presented to Sixth Graders. Four of the item types consisted of sentence-embedded PAs on the study trials, followed by test trial cues varying in terms of their similarity in meaning to the study materials. The two remaining item types served as baseline measures of performance, i.e., controls. It was found that: (1) the greater the semantic similarity between the study and test trial contexts, the greater was the recall of PA items, while at the same time; (2) control item types presented in conjunctive phrases were better recalled than those presented in sentences. Implications of these findings are discussed in terms of ongoing research into semantic and imaginal components of children's learning. (Author)