CRISP is a Monte-Carlo model of entire fire scenarios. The sub-models representing physical ‘objects’ include rooms, doors, windows, detectors and alarms, items of furniture etc, hot smoke layers, and people. The stochastic aspects include starting conditions such as windows and doors open or closed, the number, type and location of people within the building, the location of the fire and type of ... [Show full abstract] burning item. The fire and smoke spread are simulated by a 2-layer zone model. The simulation uses a variable timestep in order to achieve maximum efficiency while still maintaining a numerically stable solution.People are assumed to adopt distinct behavioural roles, either naturally or due to training. Their behaviour can be described in terms of actions, which may be abandoned, and substituted by new ones, depending on the state of the environment. Rational decisions are made based on current knowledge (which may be limited and/or incorrect).Movement of people through the building firstly requires a route to be planned through the network of rooms. The choice of route is influenced by the doors' transit difficulty (modified for the presence of smoke) and the distance. Within each room, movement to the next door on the route is directed by means of a contour map of distance to go. This enables any obstacles to be avoided. Movement speed is affected by local crowd density. Deviations from the minimum distance path through a room may be made to avoid areas of high crowd density.Some of the outputs from the CRISP model are described. These include histograms of Fractional Effective Dose (FED) acquired by the occupants (from which the life risk is calculated), and profiles of the number of people remaining in the building as a function of time. Copyright © 1999 John Wiley & Sons, Ltd.