Antidepressants have modest efficacy in late-life depression, perhaps because various neurobiological processes compromise frontolimbic networks required for antidepressant response. We propose that amyloid accumulation is an etiological factor for frontolimbic compromise that predisposes to depression and increases treatment resistance in a subgroup of older adults. In patients without history of depression, amyloid accumulation during the preclinical phase of Alzheimer’s disease may result in the prodromal depression syndrome that precedes cognitive impairment. In patients with early-onset depression, pathophysiological changes during recurrent episodes may promote amyloid accumulation, further compromise neurocircuitry required for antidepressant response and increase treatment resistance during successive depressive episodes. The following findings support the amyloid hypothesis of late-life depression: a) Depression is a risk factor, a prodrome, and a common behavioral manifestation of Alzheimer’s disease; b) Amyloid deposition occurs during a long pre-dementia period when depression is prevalent; c) Patients with lifetime history of depression have significant amyloid accumulation in brain regions related to mood regulation; d) Amyloid deposition leads to neurobiological processes, including vascular damage, neurodeheneration, neuroinflammation, disrupted functional connectivity, that impair networks implicated in depression. The amyloid hypothesis of late-life depression is timely, because availability of ligands allows in vivo assessment of amyloid in the human brain, a number of anti-amyloid agents are relatively safe, and there is evidence that some antidepressants may reduce amyloid production. A model of late-life depression introducing the role of amyloid may guide the design of studies aiming to identify novel antidepressant approaches as well as prevention strategies of Alzheimer’s disease.