PURPOSE: In eyes with intermediate age-related macular degeneration (iAMD), we separately quantified the hyperreflective foci (HRF) along the retinal pigment epithelium (rpeHRF) and the intraretinal HRF (iHRF) to determine if the location of the HRF predicted the progression from iAMD to the onset of large persistent choroidal hypertransmission defects (hyperTDs). DESIGN: Post hoc subgroup cohort analysis of a prospective study. METHODS: A retrospective analysis was performed on a prospective natural history database of eyes with AMD imaged using swept-source optical coherence tomography (SS-OCT). En face images derived from choroidal slabs positioned 64 µm to 400 µm beneath Bruch's membrane were used with a semi-automated algorithm to identify and quantify hypotransmission defects (hypoTDs) attributable to either iHRF or rpeHRF within a 5 mm fovea-centered circle. iHRF were identified on corresponding B-scans as hyperreflective lesions within the neurosensory retina, and rpeHRF were identified as areas of RPE thickening. Multivariable survival analysis was performed to determine if the area measurements of either iHRF or rpeHRF were more likely to predict the onset of the first large persistent hyperTD. RESULTS: Of the 171 eyes with iAMD included in this study, 82 (48%) developed at least one large hyperTD during a median follow-up of 59.1 months. Univariable Cox regression analyses showed that rpeHRF area (P<
0.001), iHRF area (P=0.003), and drusen volume (P<
0.001) were all significantly associated with the onset of the first large persistent hyperTD. However, a multivariable Cox regression model showed that only the rpeHRF area remained a significant predictor of disease progression (P<
0.001). CONCLUSIONS: In iAMD eyes, the area of rpeHRF was more predictive of disease progression than either the drusen volume or iHRF, which suggests that these rpeHRF serve as harbingers of focal atrophy formation and may predict where hyperTDs form.