The dependency of land use and the neighbouring land cover patches is related to the existing neighbourhood or the predominant land use types in a given region, which stimulates further land changes of the lot being observed. Land use changes can be considered a complex and (to an extent) random process. The complexity of interactions means that the conventional deductive models are constrained, particularly in the case of the spatio-temporal phenomena, and implies the application of cellular automata to spatial research. Cellular automata are similar to the map algebra models in GIS. The concept of using neighbourhood coefficients in the analysis of land use changes is based on the combination of map algebra with two-dimensional cellular automata. The aims of this research included formulating the theoretical structure of neighbourhood coefficients, analysing their operationalization and testing their practical application. The verification of research and procedures included maps and statistics of simulations of land use changes in Poland.