The effects of elaboration structure (Sentence, Semantic paragraph, Syntactic paragraph) and list length (8, 12, 16 pairs) on paired-associate learning were investigated in a 3 × 3 factorial design. Seventy-five educable retardates were tested on acquisition (S-R) and reversal (R-S) tasks. Significant acquisition differences were found in the 8-pair list, where Semantic paragraph Ss performed better than Sentence Ss. In the longer lists, all structures were equally effective in facilitating acquisition (mean first trial correct = 60%), as well as reversal (mean correct = 95%). Sentence form (declarative, imperative, interrogative) was controlled in each elaboration structure. Analyses indicated that significantly fewer acquisition errors were made on pairs presented in declarative and imperative, as opposed to interrogative, elaborations. Tests of recall for the elaborations revealed that Ss in all conditions generally recalled them as declarative sentences. Further observations at 24 pairs confirmed the 12- and 16-pair findings.