Objectives: Very few studies looking at slow gait speed as early marker of cognitive decline investigated the competing risk of death. The current study examines associations between slow gait speed and transitions between cognitive states and death in later life. Methods: We performed a coordinated analysis of three longitudinal studies with 9 to 25 years of follow-up. Data were used from older adults participating in H70 (Sweden; n = 441; aged ≥70 years), InCHIANTI (Italy; n = 955; aged ≥65 years), and LASA (the Netherlands; n = 2824; aged ≥55 years). Cognitive states were distinguished using the Mini-Mental State Examination. Slow gait speed was defined as the lowest sex-specific quintile at baseline. Multistate models were performed, adjusted for age, sex and education. Results: Most effect estimates pointed in the same direction, with slow gait speed predicting forward transitions. In two cohort studies, slow gait speed predicted transitioning from mild to severe cognitive impairment (InCHIANTI: HR = 2.08, 95%CI = 1.40–3.07; LASA: HR = 1.33, 95%CI = 1.01–1.75) and transitioning from a cognitively healthy state to death (H70: HR = 3.30, 95%CI = 1.74–6.28; LASA: HR = 1.70, 95%CI = 1.30–2.21). Conclusions: Screening for slow gait speed may be useful for identifying older adults at risk of adverse outcomes such as cognitive decline and death. However, once in the stage of more advanced cognitive impairment, slow gait speed does not seem to predict transitioning to death anymore.