Depression is associated with decreased quality of life and high mortality risk at all ages and, despite having lower prevalence in community-dwelling seniors compared to other stages of the life span or institutionalized elderly, it negatively affects seniors' health outcomes. Most evidence linking late-life depression, defined as first onset after 65 years of age, with dietary factors comes from cross-sectional studies, which cannot provide information on whether declines in diet were a consequence of depression-the reverse causality hypothesis. The Quebec Longitudinal Study on Nutrition and Aging (NuAge) provides an ideal setting to explore such a relationship among a healthy, community-dwelling elderly population. Diet is hypothesized to be a factor associated with depression. Some longitudinal evidence from studies with middle-aged adults suggests that dietary patterns such as the Mediterranean or patterns deemed "healthy" are protective against depression, while "western" and "unhealthy" patterns are risk factors for the disease. B-vitamins are important cofactors in neurotransmitter synthesis and in the enzymes of the one-carbon metabolism, which has been hypothesized to have an aetiological role in depression through its neurotoxic derivates, such as homocysteine. Nonetheless, it is still largely unknown how depression affects nutrient intake in the elderly population, raising the question of the directionality of the association. Therefore, ii the three objectives of this thesis were to explore the longitudinal associations of i) dietary patterns and macronutrient intake and of ii) B6, B12 and folate intakes with the incidence of depression in late life; and iii) to investigate whether depression significantly affects short-term nutrient intakes (reverse causality effect) in comparison to intakes among non-depressed seniors.Data for this study have been obtained from the NuAge cohort, a 4-y longitudinal study of 1,793 community-dwelling men and women aged 68 to 82 years at baseline. The main outcome is incidence of depression as defined by scores in the 30-item Geriatric Depression Scale ≥11 and or antidepressant medication use over the three years of follow-up. Participants were free of depression at the study's baseline through the exclusion of those with GDS scores ≥11 or antidepressant use at the time of the recruitment. Individuals with invalid or missing dietary information were also excluded. Dietary patterns were created through principal component analysis on amount (grams) of food items consumed in each of the 32 predefined food categories. Sex-specific tertiles of intake were created from the mean of three non-consecutive 24h-recalls. Multiple logistic regression models were adjusted for several demographic, health, and social confounders. For the study of the reverse causality hypothesis, we conducted a case-control nested within the NuAge cohort. Participants free of iii depression at baseline who developed depression at some point of follow-up were matched by age group and sex with...