Paired associate recall was tested as a function of serial position for younger and older adults for five word pairs presented aurally in quiet and in noise. In Experiment 1, the addition of noise adversely affected recall in young adults, but only in the early serial positions. Experiments 2 and 3 suggested that the recall of older adults listening to the words in quiet was nearly equivalent to that of younger adults listening in noise. In Experiment 4, we determined the signal-to-noise ratio (S/N) such that, on average, younger and older adults were able to correctly hear the same percentage of words when words were presented one at a time in noise. In Experiment 5, younger adults were tested under this S/N. Compared with older adults from Experiment 3, younger adults in this experiment recalled more words at all serial positions. The results are interpreted as showing that encoding in secondary memory is impaired by aging and noise either as a function of degraded sensory representations, or as a function of reduced processing resources. A number of recent large-scale correlational studies have shown that measures of sensory function are strongly correlated with measures of cognitive function in an aging population (Baltes & Lindenberger, 1997; Lindenberger & Baltes, 1994; Salthouse, Hancock, Meinz, & Hambrick, 1996). In the seminal study of this series, Lindenberger and Baltes showed that a hierarchical model, with five cognitive abilities (reasoning, memory, speed, knowl- edge, and fluency) as first-order factors and general intelligence as a second-order factor, provided a good characterization of the cognitive test battery. To find the best characterization of the relationship between age, sensory acuity, and these cognitive fac- tors, Lindenberger and Baltes explored a number of structural models. Surprisingly, the structural model that provided the best fit