During its earliest stage, before cell loss and independent of amyloid plaques and neurofibrillary tangles, Alzheimer's disease (AD) causes synaptic loss affecting the basal functional properties of neurons. In principle, synaptic loss can be detected by measuring AD-induced changes in basal function, or by measuring stimulus-evoked responses on top of basal changes. Functional magnetic resonance imaging (fMRI) is sensitive to both basal changes and evoked-responses, and there are therefore two experimental approaches in which fMRI can be used to pinpoint synaptic loss in AD. In this review, we will compare and contrast both approaches for pinpointing when and where synaptic loss in AD begins and for monitoring therapeutic efficacy.