The primary objective of most agricultural lield experiments and other spatial experiments is the unbiased and efficient estimation of treatment contrasts. The most common way to their estimation is to apply ordinary least squares to a linear model in which the expected value of each observation is taken to be the sum of an over all mean, a treatment effect, and various environmental effects such as block or row and column effects. This aproach yields estimators which are linear lunctions of the observations and which are unbiased if the assumed model es correct. Other alternatives, the spatial methods, have been proposed for the analysis of data from field-plot experiments. These alternatives attempt to neutralize the effect that the spatial heterogeneity of the experimental units can have on the estimation of treatment contrasts. The purpose of this paper is to analyze and compare critically some of the most important approaches about «neighbour» or «spatial» methods of analysis of field experiments, where an attempt is made to estimate and remove the effects of association of neighbouring plots from the treatment contrast.