BACKGROUND: The established relationship between vision impairment and depression is limited by the examination of depression only as a unidimensional construct. The present study explores the vision-depression relationship using a dimensional approach. METHODS: 9036 participants aged 65 years and above enrolled in the Three-City study were included. Relationships between baseline near Vision Impairment (VI) or self-reported distance Visual Function (VF) loss with trajectory of four dimensions of depression - depressed affect, positive affect, somatic symptoms and interpersonal problems - over 12 years were examined using mixed-effects models. Depression dimensions were determined using the four-factor structure of the Centre for Epidemiology Studies-Depression Scale (CESD). RESULTS: In the fully adjustment models, mild near VI predicted poorer depressed affect (b = 0.04, p = .002) and positive affect (b = -0.06, p < 0.001) over time, with evidence of longer term adjustment. Distance VF loss was associated with poorer depressed affect (b = 0.27, p ≤ .001), positive affect (b = -0.15, p = .002), and somatic symptoms (b = 0.18, p ≤ .001) at baseline, although only the association with depressed affect was significant longitudinally (b = 0.01, p = .001). Neither near VI nor distance VF loss was associated with interpersonal problems. LIMITATIONS: This paper uses a well-supported model of depression dimensions, however, there remains no definite depression dimension model. Distance VF loss was self-reported, which can be influenced by depression symptoms. CONCLUSIONS: Vision impairment in older adults is primarily associated with affective dimensions of depression. A reduction in social connectedness and ability to engage in pleasurable activities may underlie the depression-vision relationship. Older adults with vision impairment may benefit from targeted treatment of affective symptoms, and pleasant event scheduling.