Climate change is almost certain to affect snow and ice processes. Even at lower latitudes, changes in snow cover at high altitudes can significantly affect catchment hydrology. This article uses data from a transient Regional Climate Model projection (HadRM3Q0) for 1950–2099 (A1B emissions) to drive hydrological models for three nested catchments on the river Dee in north-east Scotland, to assess potential changes in flood frequency and timing using annual maxima and moving-window analyses. Some results are also shown for an upland catchment in northern England. Modelling is performed both with and without a snow module, to demonstrate the effects of snowfall/melt and how these change through time and vary between catchments. Modelled changes in flood magnitude and timing are nonlinear, with most changes for daily mean flows not significant. For longer duration (30-day) flows with snow there are significant decreases in peak magnitude, particularly for the smaller higher altitude Dee catchments, with peaks occurring months earlier in future (changes without snow are generally not significant). There is a general convergence in results with and without snow later in the period, as snow processes become less important, but convergence occurs at different times for different catchments and occurs differently for daily and 30-day peak flows due to the differential effects of snow at different durations. This not only highlights the importance of including snow processes for such catchments, particularly for longer duration flows, but also highlights the complexity of interactions: Physical catchment properties, the balance between precipitation occurrence and temperature, and how this balance alters as the climate changes will each be critical in determining the impact on the magnitude and timing of peak flows, making it hard to generalize results.