Abstract. TEX86 is a paleothermometer based on Thaumarcheotal glycerol dialkyl glycerol tetraether (GDGT) lipids and is one of the most frequently used proxies for sea-surface temperature (SST) in warmer-than-present climates. However, the calibration of TEX86 to SST is controversial because its correlation to SST is not significantly stronger than that to depth-integrated surface to subsurface temperatures. Because GDGTs are not exclusively produced in and exported from the surface ocean, sedimentary GDGTs may contain a depth-integrated signal that is sensitive to local subsurface temperature variability, which can only be proved in downcore studies. Here, we present a 15 Myr TEX86 record from ODP Site 959 in the Gulf of Guinea and use additional proxies to elucidate the source of the recorded TEX86 variability. Relatively high GDGT[2/3] ratio values from 13.6 Ma indicate that sedimentary GDGTs were partly sourced from deeper (> 200 m) waters. Moreover, late Pliocene TEX86 variability is highly sensitive to glacial-interglacial cyclicity, as is also recorded by benthic δ18O, while the variability within dinoflagellate assemblages and surface/thermocline temperature records (Uk’37 and Mg/Ca), is not primarily explained by glacial-interglacial cyclicity. Combined, these observations are best explained by TEX86 sensitivity to sub-thermocline temperature variability. We conclude that the TEX86 record represents a depth-integrated signal that incorporates a SST and a deeper component, which is compatible the present-day depth distribution of Thaumarchaeota and with the GDGT[2/3] distribution in core tops. The depth-integrated TEX86 record can potentially be used to infer SST variability, because subsurface temperature variability is generally tightly linked to SST variability. Using a subsurface calibration with peak calibration weight between 100–350 m, we estimate that east equatorial Atlantic SST cooled by ~4.5 °C between the Late Miocene and Pleistocene. On shorter timescales, we use the TEX86 record as an Antarctic Intermediate Water (AAIW) proxy and evaluate climatological leads and lags around the Pliocene M2 glacial (~3.3 Ma). Our record, combined with published information, suggests that the M2 glacial was marked by AAIW cooling during an austral summer insolation minimum, and that decreasing CO2 levels were a feedback, not the initiator, of glacial expansion.