Building on 2 previous studies (B. R. Ekstrand, 1967; B. R. Ekstrand, M. J. Sullivan, D. F. Parker, & J. N. West, 1971), the authors present 2 experiments that were aimed at characterizing the role of retroactive interference in sleep-associated declarative memory consolidation. Using an A-B, A-C paradigm with lists of word pairs in Experiment 1, the authors showed that sleep provides recovery from retroactive interference induced at encoding, whereas no such recovery was seen in several wake control conditions. Noninterfering word-pair lists were used in Experiment 2 (A-B, C-D). Sleeping after learning, in comparison with waking after learning, enhanced retention of both lists to a similar extent when encoding was less intense because of less list repetition and briefer word-pair presentations. With intense encoding, sleep-associated improvements were not seen for either list. In combination, the results indicate that the benefit of sleep for declarative memory consolidation is greater for weaker associations, regardless of whether weak associations result from retroactive interference or poor encoding.