Many concerns are raised with regard to the long-term reliability of water resources in the context of global change and,or natural climate variability. We address this issue here through a case study of the hydroclimatology of the Paraíba river basin, in Southeast Brazil. Our study shows that neither changes in the frequency and magnitude of hydrologic extremes (droughts and floods) nor changes in annual rainfall amounts can be detected in the existing climate record. Yet, a significant decrease in annual runoff production has occurred over the last 10-20 years, which cannot be traced to either urbanization or irrigation over the same period. The change is attributed to shifts in the seasonal cycle of rainfall (P) along with an increase in air temperature, especially during fall (March-April-May) and winter (June-July-August), which results in the reduction of runoff generation during the wet season (December-January-February). This hypothesis is consistent with the fact that the terrestrial water and energy cycles are tightly, and non-linearly, coupled through evapotranspiration. A hydrologic model was used to test this hypothesis by simulating the changes in hydrologic regimes of the Paraiba valley as per changes in hydrometeorological forcing before and after 1970. Results will show that small changes in rainfall timing coupled with temperature increase can have a significant impact on the basin's hydrologic regime, and thus on the availability of water resources. Often, adaptation and resiliency to climate variability are discussed in the context of catastrophic events such as floods and droughts. This study suggests that a different type of impacts, those associated with subtle changes of seasonality in the terrestrial hydrologic cycle cannot be ignored in studies of long-term sustainability of water resources.