We assessed rates of organic matter (OM) processing in coarse gravel and fine benthic sediment, along with water temperature, in four clearfell harvested and two undisturbed headwater streams flowing through wet eucalypt forest in southern Tasmania, Australia. Clearfell forestry in Tasmanian wet eucalypt forest involves felling of all timber followed by a high intensity regeneration burn to provide a receptive mineral seedbed for seedling growth. Bacterial carbon production and cellulose decomposition potential (together referred to as OM processing) were measured seasonally 3–5 years before and 2–4 years after harvesting in each stream. We employed a staircase design (staggered harvesting treatments) within a multiple before–after control–impact design to distinguish harvesting effects from natural variation. Clearfell harvesting raised the yearly mean water temperature by between 0.25 °C and 0.94 °C, and raised the maximum water temperature by between 0.84 and 1.6 °C. Rates of cellulose decomposition were not significantly correlated with sediment temperature but bacterial carbon production showed weak, significant correlations with temperature in fine (r = 0.20, P = 0.01, n = 137) and coarse gravel sediment (r = 0.39, P < 0.001, n = 137). The response in OM processing to clearfell harvesting differed between years and among benthic habitats. In coarse gravel habitat, there was a significant decrease in rates of cellulose decomposition potential in the 2nd and 4th year after harvesting, and a significant decrease in bacterial carbon production in the 3rd year after harvesting. However, we found a significant increase in rates of bacterial carbon production of fine sediment habitat in the 2nd and 4th year after harvesting. The contrasting response of OM processing between habitats indicates that habitat-specific changes occur after clearfell harvesting, which inhibit attempts to quantitatively predict downstream cumulative effects. Scaling up the habitat-specific responses will not only require estimates of the relative abundances of the distinct habitats, but may also require research into how different spatial configurations of habitats may affect reach- and catchment-scale estimates of OM processing.