Four experiments used an indirect measure of memory, perceptual identification, to examine possible differences in the generation of causal bridging and predictive inferences by young and older adults. In Experiments 1 and 2, young and older adults read paragraphs in which material necessary for understanding causal connections was either explicitly stated or merely implied. Identification of target words related to the causal connection, whether stated or implied, was better than identification of new, unrelated words. This benefit was similar in young and older adults. In Experiments 3 and 4, the generation of predictive inferences that are not necessary for discourse comprehension was examined. Overall, the results of these 2 experiments suggested no systematic age differences in the generation of these optional, elaborative inferences. Recognition memory for inferences was also tested in each experiment. The evidence suggests that causal bridging inferences, but not predictive inferences, may be incorporated into the textbase by both young and older adults. Results are discussed in terms of current theorizing about comprehension deficits in old age.