The course of sea-level fluctuations during Termination II (TII; the penultimate deglaciation), which is critical for understanding ice-sheet dynamics and suborbital climate variability, has yet to be established. This is partly because most shallow-water sequences encompassing TII were eroded during sea-level lowstands of the last glacial period or were deposited below the present sea level. Here we report a new sequence recording sea-level changes during TII in the Pleistocene sequence at Hole M0005D (water depth: 59.63 m below sea level [mbsl]) off Tahiti, French Polynesia, which was drilled during Integrated Ocean Drilling Program Expedition 310. Lithofacies variations and stratigraphic changes in the taxonomic composition, preservation states, and intraspecific test morphology of large benthic foraminifers indicate a deepening-upward sequence in the interval from Core 310-M0005D-26R (core depth: 134 mbsl) through -16R (core depth: 106 mbsl). Reconstruction of relative sea levels, based on paleodepth estimations using large benthic foraminifers, indicated a rise in sea level of about 90 m during this interval, suggesting its correlation with one of the terminations. Assuming that this rise in sea level corresponds to that during TII, after correcting for subsidence since the time of deposition, a highstand sea-level position would be 2 ± 15 m above present sea level (masl), which is generally consistent with highstand sea-level positions in MIS 5e (4 ± 2 masl). If this rise in sea level corresponds to that during older terminations, the subsidence-corrected highstand sea-level positions (30 ± 15 masl for Termination III and 54 ± 15 masl for Termination IV) are not consistent with reported ranges of interglacial sea-level highstands (− 18 to 15 masl). Therefore, the studied interval likely records the rise in sea level and associated environmental changes during TII. In particular, the intervening cored materials between the two episodes of sea-level rise found in the studied interval might record the sea-level reversal event during TII. This conclusion is consistent with U/Th ages of around 133 ka, which were obtained from slightly diagenetically altered (i.e., < 1% calcite) in situ corals in the studied interval (Core 310-M0005D-20R [core depth: 118 mbsl]). This study also suggests that our inverse approach to correlate a stratigraphic interval with an approximate time frame could be useful as an independent check on the accuracy of uranium-series dating, which has been applied extensively to fossil corals in late Quaternary sea-level studies.