A wide range of complex systems with local interactions could be described by cellular automata. Despite the fact, that in general, behaviour of each cellular automata on its own could be quite simple, their effective combination, or setting unusual interaction rules may result in extraordinary system with much more complicated behaviour, or unexpected and ambiguous observation results. Stochasticity in interactions between cells approximates simulated environment to real conditions and helps finding optimal strategy, which would be more stable under all circumstances and events, especially unlikely ones. Stochastic cellular automata are often used for modelling natural phenomena and processes, simulating traffic flows, cryptography, and so on. Finding an optimal strategy – is a key problem in managing environments with available outside influence.This article shows existence of optimal strategies for stochastic cellular automata systems, gives an example of applying improving strategy algorithm in case of extinguishing forest fires, analyses chosen strategy optimality.