It is currently unknown whether brain atrophy subtypes defined in Alzheimer's disease are clinically relevant during aging. We investigated participants (n=368) from a population-based cohort of non-demented older adults who received longitudinal neuropsychological assessments during 12 years. MRI scans at baseline and 4 years later were used to define participants with "hippocampal predominant atrophy", "cortical predominant atrophy", "homogenous atrophy" and "no evidence of brain subtype atrophy" based on the dynamics of hippocampal-to-cortical volume ratio evolution. After adjustment on age, gender, educational level and ApoE4 genotype, participants with "hippocampal predominant atrophy" declined faster regarding global cognition, verbal fluency and verbal episodic memory. In Cox proportional-hazards models, "hippocampal predominant atrophy" was associated with an increased risk of developing Alzheimer's clinical syndrome over time (HR=5.73; 95%CI 2.71-12.15), independently of age and ApoE4 genotype, the two other significant predictive factors. As a possible surrogate of confined tauopathy and early Alzheimer's disease pathology, future studies should consider the definition of "hippocampal predominant atrophy" based on hippocampal-to-cortical volume ratio evolution rather than hippocampal volume alone.