Alzheimer's disease has emerged in recent decades as a major health problem and the role of lifestyles in the modulation of risk has been increasingly recognized. Recent epidemiological studies suggest a protective effect for caffeine intake in dementia. We aimed to quantify the association between caffeine dietary intake and cognitive decline, in a cohort of adults living in Porto. A cohort of 648 subjects aged ⩾65 years was recruited between 1999–2003. Follow-up evaluation (2005–2008) was carried out on 58.2% of the eligible participants and 10.9% were deceased. Caffeine exposure in the year preceding baseline evaluation was assessed with a validated food frequency questionnaire. Cognitive evaluation consisted of baseline and follow-up Mini-Mental State Examination (MMSE). Cognitive decline was defined by a decrease ⩾ 2 points in the MMSE score between evaluations. Relative risk (RR) and 95% confidence interval (95%CI) estimates adjusted for age, education, smoking, alcohol drinking, body mass index, hypertension, and diabetes were computed using Poisson regression. Caffeine intake (> 62 mg/day [3rd third] vs. < 22 mg/day [1st third]) was associated with a lower risk of cognitive decline in women (RR=0.49, 95%CI 0.24–0.97), but not significantly in men (RR=0.65, 95%CI 0.27–1.54). Our study confirms the negative association between caffeine and cognitive decline in women.