Fire simulations and sensors are widely used in building fires,various data such as temperature,CO and CO2 concentration,visibility can be obtained by sensors and sensor-based simulation. It is important to generate a risk map based on such data so that we can use it to estimate safety of the building. In this paper,we propose a method to generate a dynamical,integrated risk map using sensor readings in a building fire. Such risk evaluation model is developed using similarity comparison between the space state and dangerous state by a likelihood distance calculating and data grouping from a two-step cluster method. The risk evaluation model considers the integrated influence on the occupants in the zone from high temperature,lack of oxygen,toxic and harmful gases and shows the relative fire risk map at certain time. Based on the simulation study,it is proved that multi-factor fire risk analysis would be more objective and accurate than single factor and two-factor risk analysis and the fire risk evaluation model can generate a risk map and provide the classification information and the whole building risk statistic results to support evacuation command and control. Fire simulations and sensors are widely used in building fires, various data such as temperature, CO and CO2 concentration, visibility can be obtained by sensors and sensor-based simulation. It is important to generate a risk map based on such data so that we can use this to estimate safety of the building. In this paper, we propose a method to generate a dynamical, integrated risk map using sensor readings in a building fire. Such risk evaluation model is developed using similarity comparison between the space state and dangerous state by a likelihood distance calculating and data grouping from a two-step cluster method. The risk evaluation model considers the integrated influence on the occupants in the zone from high temperature, lack of oxygen, toxic and harmful gases and shows the relative fire risk map at certain based on the simulation study, it is proved that multi-factor fire risk analysis would be more objective and accurate than single factor and two-factor risk analysis and the fire risk evaluation model can generate a risk map and provide the classification information and the whole building risk statistic results to support evacuation command and control.