In forestry, selective timber harvesting best management practices (BMPs) are widely adopted to mitigate the effects of clearfell harvesting on stream ecosystems. However, there have been surprisingly few studies experimentally assessing the effects of selective harvesting on freshwater benthic macroinvertebrate assemblages and how anthropogenic disturbances impact biodiversity across multiple spatial scales. We assessed how selective timber-harvesting BMPs affected alpha and beta diversity of stream macroinvertebrates when measured across three spatial scales – between harvested impact and unharvested control streams, among pools within streams, and among samples within pools. Using a multiple paired, Before-After Control-Impact design of two pairs of streams draining unharvested control and harvested impact catchments, we sampled stream macroinvertebrates bimonthly over two years before and after selective timber harvesting. Effects of selective harvesting varied between ‘impact’ streams and among different spatial scales. At the whole-stream scale, harvesting did not impact either alpha or beta diversity in both pairs of control-impact streams. Within streams, harvesting temporarily reduced beta diversity in streams with increasingly intermittent flow, yet caused brief increases in beta diversity in increasingly permanently flowing streams. At the finest spatial scale, harvesting had transient variable effects on within-pool beta diversity in one pair of experimental streams but not the other. The absence of impacts of selective harvesting on alpha diversity at all three spatial scales, and the only transient impacts on beta diversity suggest that these BMPs are sufficient to allow timber harvesting while simultaneously preserving headwater stream biodiversity. Our study provides a basis for developing monitoring programs to assess the relative impacts and consequences of anthropogenic disturbances on multi-species assemblages across different spatial scales. Our findings suggest fine scale (within-stream) beta diversity may be a more sensitive indicator of catchment disturbances than comparisons of alpha and beta diversity at broader scales among streams.