One of the most important events that cause anomalies in nature's balance is forest fires. Forest fires are also very serious threats in Turkey, which is located in the Mediterranean Region. While complete prevention of a forest fire is impossible, it is possible to reduce the damages of forest fires by constructing a forest fire risk map. In this study, we analyze a forest fire that took place at Kibriz Stream Canyon near the city of Antalya. The most important aspect of this fire was that the interference was very difficult due to the canyon's harsh landscape. Here, we show that a fire risk map can be especially beneficial for this area, and for other areas where fire interference is very difficult due to the unfriendly topography. Remote Sensing and Geographic Information System (GIS) were applied in this study for the assessment of the situation before and after the fire and for forming the fire risk map. Landsat TM (01.08.1990), Spot XS (24.06.2007) imageries and 1/25000 scale topography maps were used to generate a digital terrain model and to establish the land use classes by means of unsupervised and supervised classification algorithms. In addition, the normalized difference vegetation index (NDVI) was computed to compare the classified imageries before and after the fire. Comparison of the NDVI values helped us to determine the vegetation pattern change after the fire.