Climate variability might temporarily improve or mitigate the effects of increasing global warming. Understanding and estimating internal variability is just as important as understanding the role of anthropogenic forcing, as the combination of both drives climate events in the real world. The objective of this work is to analyze the relationship between the SST of the equatorial Pacific and 4 extreme temperature indices in southern South America considering gridded observational data (HadEX3), reanalyses (ERA-Interim, NCEP1, NCEP2), and global climate models participating in the Coupled Model Intercomparison Project Phase 5 (CMIP5) in the historical period 1979-2005. For this, correlations and quantile regression for the 90th percentile were estimated between the variables. Moreover, to assess the performance of the reanalysis and CMIP5 models, multiple metrics were calculated. The observations showed that warm conditions in the equatorial Pacific are mainly associated with a higher occurrence of warm nights in the north and center of Argentina and Chile in winter and spring. However, the different reanalyses considered in this study showed discrepancies in representing these relationships. Several CMIP5 models were generally able to simulate correlation patterns for the warm extremes of the minimum and maximum temperature in comparison to HadEX3. These types of studies are critical to understanding whether climate models simulate temperature extremes in association with physical processes, providing greater confidence in their future projections.