A 6-year study documented the effects of clear-cut harvesting with and without riparian buffers (10 m and 30 m wide) on headwater stream temperature in coastal British Columbia. The experiment involved a replicated paired catchment design. Pretreatment calibration relations between the treatment and control streams were fitted using time series of daily minimum, mean, and maximum temperatures. Generalized least squares (GLS) regression was used to account for autocorrelation in the residuals. While water temperature in streams with 10 and 30 m buffers did not exhibit marked warming, daily maximum temperature in summer increased by up to 2¿--8¿C in the streams with no buffer. The effectiveness of the buffers may have been maximized by the north-south orientation of the streams, which meant that the streams would be well shaded from late morning to early afternoon by the overhead canopy, even under the 10 m buffer. The variation in response for the no-buffer treatments is consistent with the differences in channel morphology that influence their exposure to solar radiation and their depth. Relations between treatment effect and daily maximum air temperature suggested that recovery toward preharvest temperature conditions was occurring, with rates appearing to vary with stream and by season.