Using Landsat imagery, this study was conducted to evaluate a fire disturbance that occurred in Canada’s Grasslands National Park on 27 April 2013. We used spectral indices (e.g. Normalised Burn Ratio (NBR) and Mid-infrared Burn Index (MIRBI)) derived from Landsat images to evaluate burn severity and to analyse the vegetation recovery process. A field survey was conducted to assess burn severity, which we used to evaluate the performance of spectral indices. Responses of the vegetation community to the fire disturbance were also investigated during the field campaign. Results show that the selected spectral indices performed differently for evaluating burn severity, but MIRBI performed best, likely due to its ability to discriminate post-fire residuals. Severely burned areas were distributed along a river where a larger amount of senesced biomass had accumulated before the fire. The semiarid grasslands showed a strong resilience to fire disturbance, and vegetation recovery was likely influenced by burn severity and water availability. Different vegetation types (e.g. grass, trees and shrubs) had distinct recovery rates and, thus, fire influences plant community development. The fire disturbance changed the composition of grass species in the burned area and also promoted invasion by non-native species.