Transfer-appropriate processing theories differentiate between conceptual-and perceptual-priming tasks. The former are said to be influenced by the nature of processing engaged in at study, but not by changes in modality between study and test; the latter are sensitive to changes in format between study and test, but not to variations in the extent ofsemantic processing at study. In the present experiments, we examined the effects of divided attention and aging on priming in exemplar generation and category verification, two tasks that require access to semantic information at test. Manipulations of attention during encoding affected the extent of priming in exemplar generation, but not in category verification. Priming effects were similar in young and older adults in exemplar generation following study in both full and divided attention. Although older adults did not demonstrate priming in category verification in one experiment, no effects of age or divided attention were observed in a second experiment. In addition, priming in category verification was unaffected by varying the level of processing at encoding. However, the absence of levels-of-processing and attention effects in category verification does not signal that priming in this task has a perceptual basis; priming in category verification was insensitive to modality shifts between study and test. The implications of these fmdings for theories of priming and cognitive aging are considered.Division of attention during encoding is well known to have negative consequences for performance on direct measures of memory, such as recall and recognition, that involve deliberate recollection of prior study episodes (e.g, Craik, Govoni, Naveh-Benjamin, & Anderson, 1996;Fisk & Schneider, 1984;Kellogg, 1980;Parkin, Reid, & Russo, 1990;Weldon & Jackson-Barrett, 1993). However, the consequences of dividing attention during encoding on indirect memory tasks, in which the task instructions do not mention the relation between the task to be performed and the prior study episodes, are less well understood. It is, for instance, unclear whether we should expect the effects of dividing attention to be uniform across all classes of indirect memory tasks.Recent research on the consequences of dividing attention for indirect measures of memory has been guided largely by the transfer-appropriate processing frame-