Although temperature has been identified as a vital parameter within the aquatic environment, relatively little research has been undertaken into how temperature can vary within rivers and streams during storm periods. This study examines water temperature variations during such events at four sites within the Middle and Upper Exe basin, in South-West England. Statistical analysis is used on longer-term data to assess the importance of discharge levels in determining water temperature, at several temporal scales. Isolation and examination of all storm periods recorded within a 20-water year data set identifies and classifies the forms of water temperature modification occurring. Various hydrometeorological and flow parameters are used to explain the observed patterns of water temperature behaviours during storm events, using both existing data and new information gathered from a network of data loggers, and a storm sampling campaign. Additionally, an assessment of the use of water temperature as a tracer of runoff source during storm periods is considered. Instrumentation of two contrasting small catchments within the Exe basin provided high quality on-site hydrometeorological data, for the 1998-99 water year. These data are used to reinforce the classification system proposed, via both specific case studies of individual periods and statistical analysis of all the storm events. Likely causes of the observed patterns are identified, with rain intensity and time to peak discharge being particularly significant. An intensive sampling campaign was carried out to further justify the classification of events, with spot-sampling of runoff sources throughout storm-periods. The spatial variability of temperature within these sources is discussed. The use of stream temperature as a tracer of runoff source is investigated, in comparison with geochemical properties. It was found that at the micro-scale,u sing water temperature was at least as useful as using more traditional chemical tracers. The study finds that a large percentage of storm events modify stream water temperatures, most commonly by modifying the diel range of temperatures in the channel, in comparison to preceding dry periods, but also presents examples of events where sudden short-lived thermal changes are present, related to thermally distinct inputs.