Using a reciprocal exchange experimental design, the processing rate of hazel leaf packs was measured in two low-order streams in south-west Ireland that differed in their macroinvertebrate assemblages and in their susceptibility to acid pulses during spates. One stream had a deciduous riparian zone and the other flowed through a conifer plantation. The influence of the detritivore assemblage relative to the initial microbial conditioning was assessed. Hazel leaf packs and plastic 'habitat-only' packs were initially conditioned in each stream and were then either transferred to the other stream or replaced in the stream of conditioning. The rate of macroinvertebrate colonisation and hazel leaf pack breakdown were subsequently measured in all four treatments. There was no significant difference between streams in leaf mass loss during the conditioning period. During the subsequent exposure period, there was a significant difference between streams in hazel leaf breakdown rate, regardless of the stream in which the leaf pack was conditioned. Within each stream, detritivores (shredders and deposit feeders) did not differentially colonise packs conditioned in the different streams, indicating that there was no consequential difference in initial conditioning between streams. However, there were significant differences between the streams in their colonising detritivore assemblages, both in the abundance of individual taxa and in the relative dominance of different functional feeding groups. From this study it would appear that the impact of detritivore activity in the leaf packs outweighs any potential differences in microbial processing between the two streams and also exceeds the effect of differential conditioning on the attractiveness of the detritus. Factors influencing the structure of the detritivore assemblages are thus of primary importance in the decomposition of leaf litter in these systems.