Probabilistic techniques offer rational methods of dealing with the randomness of fire risk and effectiveness of firesafety measures. In this paper the state of the art in these techniques is reviewed with examples illustrating the use of fire statistics in evaluating the parameters of these models. Methods discussed include exponential model of fire growth, probabiltiy distribution of damage, and stochastic model of fire spread. The estimates provided by these models can be used in conjunction with an engineering method for a probabilistic treatment of deterministic predictions.