Depression is a known risk factor for dementia. Antidepressants are the most commonly used treatment for this condition, and are effective in at least half to two-thirds of cases. Extensive evidence from in vitro and animal models suggests that antidepressants have anti-inflammatory and neuroprotective properties. These effects have been shown to reduce the oxidative damage, amyloid aggregation, and expression of pro-inflammatory genes associated with animal models of neurodegenerative disorders. However, longitudinal research in humans has shown that antidepressants do not protect against dementia, and may even be associated with a risk of cognitive deterioration over time in older adults. The contrast between two sets of findings represents a paradox of significant clinical and public health significance, particularly when treating depression in late life. This review paper attempts to resolve this paradox by critically reviewing the medium- and long-term effects of antidepressants on peripheral immune-inflammatory responses, infection risk, gut microbiota, and neuroendocrine responses to stress, and how these effects may influence the risk of neurodegeneration. Briefly stated, it is possible that the peripheral actions of antidepressant medications may antagonize their beneficial effects against neuroinflammation. The implications of these findings are then explored with a particular focus on the development and testing of multimodal neuroprotective and anti-inflammatory treatments that could reduce the risk of Alzheimer’s and related dementias in patients suffering from depression.