Young and older adults provided language samples in response to questions while walking, finger tapping, and ignoring speech or noise. The language samples were scored on 3 dimensions: fluency, complexity, and content. The hypothesis that working memory limitations affect speech production by older adults was tested by comparing baseline samples with those produced while the participants were performing the concurrent tasks. There were baseline differences: Older adults' speech was less fluent and less complex than young adults' speech. Young adults adopted a different strategy in response to the dual-task demands than older adults: They reduced sentence length and grammatical complexity. In contrast, older adults shifted to a reduced speech rate in the dual-task conditions.