Background. Within-person growth in life satisfaction (LS) can protect against declines in cognitive functioning, and, conversely, over time. However, most studies have been crosssectional, thereby precluding causal inferences. Thus, we used bivariate dual latent change score modeling to test within-person change-to-future change relations between LS and cognition. Method. Community adults completed in-person tests of verbal working memory (WM), processing speed, spatial cognition, and an LS self-report. Five waves of assessment occurred across 23 years. Results. Reduction in LS predicted future decreases in spatial cognition, processing speed, and verbal WM (|d| = 0.150-0.354). Additionally, depletion in processing speed and verbal WM predicted a future decrease in LS (d = 0.142-0.269). However, change in spatial cognition did not predict change in LS (|d| = 0.085). Discussion. LS and verbal WM and processing speed predicted one another across long durations. Evidence-based therapies can be augmented to target LS and cognition.