In the context of climate change and human-induced habitat destruction, biodiversity is being destroyed at an unprecedented rate. Moreover, the biodiversities in freshwater ecosystems are decreasing faster than those in terrestrial ecosystems. Less attention, however, has been paid to this issue. Compared with fish and algae, benthic macroinvertebrates, which have more advantages, become the most popular biological indicators of river health. Early studies were mostly focused on the influence of small-scale indicators on macroinvertebrates assemblages. Since the 1990s, some researchers started to pay attention to the impact of large-scale factors. Factors at landscape and watershed scale began to be treated equally or were given more concern. Understanding the relative influence of environmental variables at different scales is important in the distribution of large-scale monitoring points and river restoration. Based on environmental and biotic data of 38 sites within the Luan River Basin, the aims of the study were to identify the key factors related to macroinvertebrates at reach and catchment scales, and to determine which scale is more important in explaining the variation of macroinvertebrates assemblages in this region. Macroinvertebrate assemblages were collected by D-frame net and described by using community structure index and biotic attributes. Such methods were commonly used to describe ecological condition of streams (e. g., richness, feeding type). Water chemistry and physical indicators were involved at reach scale. A majority of reach-scale physical indicators were measured using the method in reference to rapid biological assessment programs in the United States and the Australian community monitoring manual. The covered ratio of forest (%), grassland (%), arable land (%) and construction land (%) in each catchment and buffer zone were summarized as catchment-scale indicators. Redundancy analyses (RDA) was used to quantify macroinvertebrate-reach relationships. Key variables for each ordination at reach scale were selected using the manual forward-selection procedure provided in CANOCO 4.5. The selection was based on Monte Carlo permutation test. Then, the RDA was performed to test for the significance of the first RDA axis and all axes, only by the selected reach-scale variables. The following results were captured as follows. First, 117 macroinvertebrate taxa were collected. The dominant population community was aquatic insects which included 107 generas. Second, RDA results showed that the selected reachscale variables related to macroinvertebrates were %fine sediments, riparian vegetation cover, riparian human disturbance, % cropland in riparian, water width and river transformation. The first axis and all axes explained 0.307 and 0.42, respectively. Both reached a significant level. The first axis and riparian vegetation cover had a highest correlation coefficient. Third, the selected catchment-scale variables were latitude, attitude, % cropland in the whole catchment and catchment area. The first axis and all axes explained 0.234 and 0.32, respectively. Both of them also reached a significant level. The first axis and catchment area had a highest correlation coefficient. The results showed that benthic macoinvertebrate communities in Luanhe River Bsin were affected by the combined effects of natural and human factors at both scales. However, variables at reach scale played a more important role in stream macroinvertebrates distribution, rather than catchment scale. It was suggested that more attention should be paid to reach-scale factors in river ecosystem management and restoration in Luanhe River Basin.