A robust understanding of past oceanographic variability in the Southern Ocean is important because of its role in modulating global climate change. Here, we analyzed the distributions of isoprenoid glycerol dialkyl glycerol tetraethers (GDGTs), both non-hydroxylated and the more recently discovered hydroxylated ones, in a well studied 500 kyr sediment record (core PS2489-2) from the Atlantic sector of the Southern Ocean and reconstructed past sea surface temperature. Given the uncertainty in the GDGT temperature indices, we appraised existing calibrations by comparing them with other temperature proxies and cold-water mass indicators determined from the same core. None of the existing calibrations afforded temporal trends and/or absolute values consistent with other better constrained temperature proxies. Using an extended compilation from a global core top hydroxylated GDGT data set, we examined if the disagreement might stem from the calibration data set and the definition of the GDGT indices. Among the new GDGT indices tested, the OHC index (an extended TEX86 index modified similarly to the UK37 index) and OHL (including a log function similar to TEX86L) showed temporal variability that was the most consistent with other proxies. However, they also gave unrealistic sub-zero glacial temperature values, which may have been caused by a biased calibration due to the small calibration data set, and/or a shift in production or export depth of GDGTs during glacial stages which, in turn, result in a GDGT-temperature relationship different from that during the interglacial stages.