Acquisition of the words this and thatwas tested by a comprehension and a production task with children of ages 4 and 7. In the comprehension task, children had to choose one of two toys requested by the experimenter. Child and experimenter were seated either side by side or facing one another. It was expected that children would first learn the spatial contrast between the two words and then learn to apply it from the speaker's point of view, according to developmental stages similar to those found in other studies of spatial non-egocentrism. The prediction was confirmed, but comparison with the results of de Villiers & de Villiers (1974) suggests an alternative theoretical analysis, in which non-egocentrism is not necessary for correct use of this and that. A second prediction, that performance would be better on the unmarked word that, was not confirmed for the side-by-side condition. In the production task, only 2 of the 60 children violated the restriction that this can refer only to near objects.