Extreme hydroclimate events affect the thermal regimes of freshwater systems. In a four-year period spanning periods of flooding and severe drought in Southeastern South America (2019-2023), we monitored temperature variability in streams and floodplain lakes in an undammed tributary of the Lower Uruguay River, using Onset HOBO dataloggers and modeled daily streamflow (Q). We compare the performance of linear and nonlinear regression and generalized additive models (GAM) to evaluate how air temperature (T air) variability affects water temperature (T water) at daily, weekly and monthly time scales and under different Q percentile ranges. Mean daily temperatures ranged from 6.4ºC to 30.2ºC with daily maxima coinciding with record highs >40 ºC in T air and record low streamflow in summer months. Daily and weekly air-water relationships for rivers, streams and lakes were best fit by nonlinear sinusoidal models (Nash Sutcliffe Error - NSE ≥ 0.68). Extremely low-flow periods (<10% Q) showed a linear relationship to T air, while high-flow periods (>90%) showed a nonlinear sinusoidal air-water relationship. Moreover, high-flow events showed weaker T air-T water model perforamce than moderate to low flow periods. Preliminary comparisons between basin land-cover and T water suggest that forest cover favors lower T min. This study fills a geographical gap in thermal regime data in freshwater ecosystems in South America and suggests that extreme hydroclimate events have important implications for the thermal behavior of freshwater systems.