Children in Grades 2 and 5, young adults, and older adults generated a personal word association to each of 24 target items prior to a surprise cued-recall test. The cues were either the subject-generated associates or category labels. Subjects at all ages recalled more associate-cued than category-cued items. Using the performance of the young adults as a baseline, the second-grade children performed equally poorly with both types of cues, the fifth-grade children performed relatively more poorly with category than associate cues, while the older adults performed relatively more poorly with associate than category cues. The results are discussed in terms of different patterns of age change during childhood and adulthood in the tendencies to form contextually distinctive encodings and to encode category information spontaneously. RESUME Des enfants de niveau 2 et 5, des jeunes adultes et des adultcs plus vieux produisaient une association personnelle de mots a chacun de 24 items cibles avant ('introduction d'un test de rappel surprise avec indices. Les indices etaient soit les associations produites par le sujet soit des etiquettes de categoric Les sujets de tout age se rappelaient plus d'items indices-associes qu'indices-categorie. Utilisant la performance des jeunes adultes comme point de reprere, les enfants de niveau 2 presentaient unc performance aussi pauvre avec les deux types d'indices, les enfants de niveau 5 etaient relativement plus mauvais avec les indices categorie qu'associes, alors que pour les adultcs plus vieux ce fut I'inverse.The ability to memorize and recall information improves dramatically during the elementary school years, reaches a peak during young adulthood, and declines in old age. Much of the change results from age differences in the use of deliberate mnemonic strategies such as rehearsal, organization, and elaboration (Craik, 1977;Guttentag, 1985;Kail, 1984; Salthouse, 1982). However, age differences across the life-span have also been found on incidental memory tasks (Craik & Simon, 1980; Ghatala, Carbonari, & Bobele, 1980; Perlmutter & Mitchell, 1982), a finding which likely reflects changes with age in information encoding.Craik (Craik & Byrd, 1982;Rabinowitz, Craik, & Ackerman, 1982) has proposed that age differences during adulthood in incidental memory performance reflect a reduced tendency on the part of older adults to process to-bememorized information in a distinctive, contextually-specific fashion. Instead,