*Correspondence to: Sherri L. Johnson, Forestry Sciences Laboratory, 3200 SW Jefferson Way, Corvallis, OR 97331, USA. E-mail: sherrijohnson@fs.fed.us Stream temperature dynamics, which influence most in-stream biological processes, have been the focus of much controversy. Stream temperatures have become a major issue in some regions and the centre of policy debate, because elevated temperatures can negatively impact cold-water fish species, such as threatened or endangered salmonids. To add to the controversy, numerous contradictions exist in published literature about the controlling factors of stream temperature: the role of air temperature (Sullivan and Adams, 1990; Webb and Nobilis, 1997); effects of shade (Larson and Larson, 1996; Beschta, 1997); substrate fluxes and conduction (Brown, 1969; Webb and Zhang, 1997); and changes in longitudinal temperature trajectories downstream of harvested areas (Beschta et al., 1987; Zwieniecki and Newton, 1999). Although the influences of stream temperature appear to be simple, we have much to learn about these complex processes. This commentary addresses two main points regarding modelling of stream temperature dynamics: the problems with air temperature correlations to predict stream temperature and the importance of scaling of factors, both microclimatic influences as well as reach-scale upstream influences. An important distinction in understanding the influences of stream temperature is the differences between correlation and causation. For example, high correlations exist between air and stream temperature in the diurnal and seasonal patterns of temperature fluctuations; air temperature is occasionally used as a ‘surrogate’ for predictions instead of complex heat flux equations (Webb, 1987). Unfortunately, the literature has numerous examples of statements that air temperature is a major driver of stream temperature (Smith and Lavis, 1975; Sullivan and Adams, 1990), which can be confusing to readers unfamiliar with stream temperature dynamics. Heat budget analyses show that convection, or the transfer of heat energy from warmer air to cooler stream water is, in fact, only a small portion of the energy exchanges influencing stream temperature (Sinokrot and Stefan, 1993; Webb and Nobilis, 1997). The major factor influencing both air and stream temperature is incoming solar radiation. Correlations can be helpful in predicting patterns for a future time or a nearby location, but correlations do not imply causation. Complex environmental gradients occur over very short distances away from the stream. Wind speed, relative humidity, subsurface saturation and soil and air temperature are very responsive to