Developmental aspects of relearning were examined in 3-year-olds, 5-year-olds, and college students by using Nelson's (1971) revised savings paradigm. A set of picture pairs was learned to a recognition criterion of one error-free trial. After 2 weeks, the subjects returned for tests of retention and relearning. During relearning, the subjects received one study/test trial of a set composed of pairs seen previously ("old pairs") and pairs created by re-pairing pictures seen previously ("new pairs"). Significant savings was found for all age groups on old pairs relative to new pairs. A developmental trend toward decreased savings with age was noted. Results are discussed with respect to the relative contributions of savings and practice to retention.Much of the interest in cognition and cognitive development has been directed at understanding the processes of memory. Typically, memory is studied by asking subjects to learn some information and then by testing their ability to either recognize or recall that information at a later time. Consequently, much is known about changes in recognition and recall with age (e.g., Cole, Frankel, & Sharp, 1971;Flavell, Beach, & Chinsky, 1966). In contrast, Ebbinghaus's (1885/1913) relearning method has been used infrequently by developmentalists, therefore little is known about developmental differences in savings.Relearning assesses the amount of time or effort one "saves" in (re)learning material that was acquired at an earlier time but is presently inaccessible. Although the original procedure suffered from several methodological flaws (see Bunch, 1941;Duncan, 1960;and Luh, 1922, for specific criticisms), a revised version has remedied many of its shortcomings (Nelson, 1971(Nelson, , 1978(Nelson, , 1985. In particular, Nelson has dealt with the criticism that relearning is confounded with learning-to-Iearn, or the "practice effect." Ebbinghaus's (1885/1913) original procedure, which compared the learning of a serial recall list with the relearning of that same list, made practice effects quite probable. In the revised version, pairedassociate tasks are used, and comparisons of original learning and relearning are made within a single session. That is, relearning of pairs acquired _ previously ("old pairs") is compared with learning of pairs that are comparable, but not identical, to old pairs ("new pairs"). This manipulation makes it possible to separate practice effects from savings and, thus, to more clearly distinguish beThis article is based on a doctoral dissertation written by the first author under the direction of the second author. We are grateful to the children, parents, and teachers of the Colorado State University Daycare Center, Daybridge Learning Centers, Jack & Jill Daycare, and United Daycare in Fort Collins, Colorado, for their participation in the study. Requests for reprints should be sent to Marilyn Livosky, Department of Psychology, Behrend College, Penn State Erie, Erie PA 16563-{)500. tween learning and memory. It is this revised version that has bee...