Objectives To investigate whether gait speed predicts incident depressive symptoms and whether depressive symptoms predict incident gait speed impairment; to ascertain the presence of shared risk factors for these associations. Design The Longitudinal Aging Study Amsterdam, a prospective cohort study with five follow-up cycles over 16 years. Setting Population based. Participants One thousand nine hundred twenty-eight respondents for incident depressive symptoms (mean age 68.9 ± 8.5) and 1,855 respondents for incident gait speed impairment (mean age 68.0 ± 8.2). Measurements Depressive symptoms were measured using the Center for Epidemiologic Studies Depression Scale; gait speed was measured, back and forth, during a 3-m walk as quickly as possible, with a 180° turn. Multivariate analyses were performed for both sexes using Cox regression. Results Incident depressive symptoms occurred in 24% of respondents. In univariate analyses, gait speed at baseline predicted incident depressive symptoms in men and women; after adjustment for covariates, this association persisted in men only. Examining the reverse association, 34% of respondents developed gait speed impairment. Depressive symptoms at baseline were univariately associated with incident gait speed impairment in women but not in men; this association did not persist after adjustment. The bidirectional associations did not share the same explanatory variables. Conclusion Gait speed predicts depressive symptoms in men. The geriatric giants of depressive symptoms and slowed gait speed in late life appear to result from different pathologies, both of which therefore require their own treatment strategies. © 2012, Copyright the Authors Journal compilation © 2012, The American Geriatrics Society.