The reconstruction of accurate sea‐surface temperatures (SST) is of utmost importance due to the ocean’s central role in the global climate system. Yet, a number of environmental processes might bias reliable SST estimations. Here, we investigate the fidelity of SST reconstructions for the western tropical South Atlantic (WTSA) for the interval covered by Marine Isotope Stages (MIS) 6–5, utilizing a core collected off eastern Brazil at ∼20°S. This interval was selected as previous SST estimates based on Mg/Ca ratios of planktic foraminifera suggested a peculiar pooling of warm surface waters in the WTSA during MIS 6 despite glacial boundary conditions. To ground‐truth the Mg/Ca‐based SST data we generated SST reconstructions on the same core material using the alkenone and TEX86 paleothermometers. Comparison with alkenone‐based temperature estimates corroborate the previous Mg/Ca‐based SST reconstructions, supporting the suggestion of a warm‐water anomaly during MIS 6. In contrast, TEX86‐derived temperatures, albeit representing annual mean SST in recent core top samples, are up to 6°C colder than Mg/Ca‐ and alkenone‐based SST reconstructions. We interpret the periods of anomalously cold TEX86‐temperatures as a result of a vertical migration of the TEX86 producers (heterotrophic marine Thaumarchaeota) towards greater water depths, following food availability during phases of enhanced fluvial suspension input. Likewise, the data suggest that alkenone‐based SST are, albeit to a minor degree when compared to TEX86, affected by river run‐off and/or a seasonal bias in the growth season of haptophyte algae.