Cattle are responsible for the deterioration of aquatic and riparian ecosystems throughout the North American prairies. Marked preference for riparian areas has resulted in vegetation loss, stream bank destabilization, changes in sediment particle size, and increased nutrient loads in the streams. A grazing experiment in Grasslands National Park of Canada manipulated the density of cattle to represent a range of grazing intensities (from no grazing to very heavy grazing, 70% forage utilization). This experiment provided the opportunity to study how streams in the semi-arid mixed-grass prairie environment respond to a range of grazing pressure. Nine experimental pastures located on previously ungrazed land within the park boundary and four located within the adjacent community pastures were created, each subjected to a specified grazing treatment. Sampling occurred in the autumn from 2007 to 2009 and included the measurement of 33 physical, chemical, and biological habitat metrics and the characterization of the aquatic invertebrate community. Linear regressions were performed to determine if the habitat variables had a significant relationship to grazing intensity (P < 0.05). Of the habitat variables, two sediment particle size categories were significantly related to grazing intensity: per cent of fine gravel (4-8 mm diameter) (P =0.003) and per cent of medium gravel (8-16 mm diameter) (P = 0.007). The only other habitat variable with a significant linear relationship to grazing intensity was the concentration of suspended carbon in the stream water (P = 0.050). Three invertebrate community metrics were focused on for their expected response to changes associated with cattle impacts: iii per cent Chironomidae (%Chiron), per cent Ephemeroptera, Odonata, and Trichoptera (%EOT), and taxa richness. There was a significant non-linear relationship between %Chiron (P = 0.005) and grazing intensity, no linear or non-linear relationship between %EOT and grazing intensity, and a non-linear trend between richness and grazing intensity (P = 0.083). A Reference Condition Approach was used to test for the effects of grazing on the invertebrate community. Multiple regression was used to create a model predicting the invertebrate community from habitat metrics. Of the three community metrics, only richness resulted in a model with acceptable predictive ability. The predicted richness values for each test site were calculated and their residuals were determined and compared to the distribution of residuals observed in the reference sites. Using this technique, I determined that 73.3% of the sites subjected to grazing deviated significantly for the reference condition and were therefore deemed to be impacted. There was no significant relationship between the test site residuals and grazing intensity. The macroinvertebrate community in this semi-arid environment is already under a lot of stress, the addition of cattle to the environment, even at low intensities, pushed the community beyond the reference condition.