In this paper, we investigated the influence of contrasting riparian forest types on upland stream temperatures and the links between thermal regime and standard indices of forest cover. Stream temperature was measured for two hydrological years at multiple sites along two tributaries of the River Dee, Scotland. Riparian cover in the upper reaches of both streams was dominated by Calluna moorland while the lower sections comprised either dense alluvial broadleaved forest (Girnock Burn) or native Scots Pine stands (Tanar). The Point-Centered-Quarter Method was used to quantify stem density (SD) and tree coverage (TC) as an indicator of channel shading comprising either dense alluvial broadleaved forest (Girnock Burn) or native Scots Pine stands (Tanar). The greatest reductions in maximum temperature (4 °C) and diel amplitude (6·6°C) were observed where TC and SD were greatest in areas of relatively dense broadleaved riparian cover. Analysis of the relationship between forest characteristics and temperature summary statistics revealed that SD and TC were significant predictors of mean and maximum temperatures, as well as daily temperature ranges for both rivers during the summer. The relationship between summer thermal indices and indicators of channel shading (SD or TC) was independent of forest type. However, as SD and TC were greater in the deciduous reaches of Glen Girnock than the mature Scots pine reaches of Glen Tanar, maximum reductions in stream temperature were greater in Glen Girnock. Given that maximum temperatures were strongly influenced by measures of shading, it suggested that riparian planting may offer a suitable mitigation against high temperatures under climate change. Copyright © 2012 John Wiley & Sons, Ltd.