Young and older adults performed a memory search task in which, before probe onset, a cue indicated which of 4 memory-set items the probe was most likely to be. The results were consistent with an attentional allocation model in which performance represents a weighted combination, across trials, of focused (i.e., selective) versus distributed attention. The model significantly underestimated the reaction time required by miscued trials, probably because of the response inhibition occurring on these trials. The degree to which Ss relied on focused attention was significantly greater for older adults than for young adults. The estimated time required to shift attention between memory-set items was equivalent for the 2 age groups.