Studies of sea level during previous interglacials provide insight into the stability of polar ice sheets in the face of global climate change. Commonly, these studies correct ancient sea-level highstands for the contaminating effect of isostatic adjustment associated with past ice age cycles, and interpret the residuals as being equivalent to the peak eustatic sea level associated with excess melting, relative to present day, of ancient polar ice sheets. However, the collapse of polar ice sheets produces a distinct geometry, or fingerprint, of sea-level change, which must be accounted for to accurately infer peak eustatic sea level from site-specific residual highstands. To explore this issue, we compute fingerprints associated with the collapse of the Greenland Ice Sheet, West Antarctic Ice Sheet, and marine sectors of the East Antarctic Ice Sheet in order to isolate regions that would have been subject to greater-than-eustatic sea-level change for all three cases. These fingerprints are more robust than those associated with modern melting events, when applied to infer eustatic sea level, because: (1) a significant collapse of polar ice sheets reduces the sensitivity of the computed fingerprints to uncertainties in the geometry of the melt regions; and (2) the sea-level signal associated with the collapse will dominate the signal from steric effects. We evaluate these fingerprints at a suite of sites where sea-level records from interglacial marine isotopes stages (MIS) 5e and 11 have been obtained. Using these results, we demonstrate that previously discrepant estimates of peak eustatic sea level during MIS5e based on sea-level markers in Australia and the Seychelles are brought into closer accord.