The study presents the results obtained following the analysis of the so-called Granger causality between daily and monthly temperatures of air and water for the period 1987–2013 carried out for the Noteć river and its two main tributaries: Drawa and Gwda. Granger causality relates to a situation where the data concerning past values of one time series provide important information helping to predict values of another series not included in the information about its past values. The analysis was based on the Granger causality test (of the first order). A causality relationship was established for daily temperature series both for the air-water and water-air directions of influence, which means that forecasting the pattern of river water temperatures from changes to air temperatures can yield better results when done based on data from the previous day. The model forecasting daily water temperature in the Noteć river on the basis of water and air temperatures from the previous day explained 0.07–0.27% of unique variance more than the model that used only water temperature from the previous day. The model forecasting the daily air temperature based on air and water temperatures from the previous day explained 0.3–0.79% of the variance more than the model, which uses only the air temperature from the previous day. For monthly series of water and air temperatures, different configurations of correlations in terms of Granger causality were established: one-way in water-air direction or no correlation, which may result from the river water thermal regime being disturbed by the local impact of anthropogenic factors. In addition, the analysed effect of Granger causality between series of random fluctuations of both temperature models confirmed that causal dependencies occur in both directions. The identification of causal relationships in terms of Granger causality confirms the influence of one data series on the evolution on another data series, and it defines the application potential of study results for the purpose of forecasting the changeability of thermal parameters of river waters. The obtained results may help improve the quality of forecasting changes in water thermal conditions, which is important for managing their environmental condition properly.