Maintaining the natural complexity of water temperature regimes is a key to maintaining diverse biological communities. Insect communities, food webs, and fish respond to the magnitude and duration of water temperature fluctuations. Disruption of these natural patterns has the potential to alter physiological processes, behavioural adaptations, and community structure and dynamics. We analysed multiple >300-day time series of water temperature from the Willamette River basin, Oregon, to assess the impact of large multi-purpose dams on water temperature variability at temporal scales ranging from 1 to 32 days, short temporal scales that are commonly ignored. We applied wavelet analysis to quantify the variability of water temperature at multiple temporal scales simultaneously. We compared water temperature regimes above and below dams and before and after dam construction. The advantages of wavelet analysis are the ability to examine all temporal scales simultaneously and independently as well as the ability to preserve the temporal context of the wavelet coefficients. We were able to detect significant (p < 0.0001) reductions in water temperature variability, defined as the variability of the wavelet coefficients, as a result of dams at the 1-, 2-, 4-, and 8-day scales. There were no significant differences in water temperature variability between managed and natural flows at the 16- and 32-day scale (p = 0.80). In addition to the well-documented effects of dams on seasonal patterns in water temperature or on water temperature extremes, our results demonstrate that dams have significantly muted the small temporal scale variance in water temperature patterns to which many organisms may have been adapted. Conserving or restoring natural temperature patterns in rivers will require attention to these small-scale complexities. Published in 2007 by John Wiley & Sons, Ltd.