It has been shown that discriminating the orientations of a target is influenced by the spatial distribution of the orientations of similar elements in the background. However, this effect appears to be essentially decisional (Dufour and Bonnet, 1995 Spatial Vision9 307 – 324). In the present experiment, we explored the accuracy with which subjects can discriminate relative proportions of orientations distributed over a surface (background). Stimuli were textures made of 100 segments with regular spacing. Each of the segments had one of four possible orientations. For each display, one of the orientations was overrepresented (31%, 37%, 43% and 49%). The task of the subject was to discriminate, in a reaction-time paradigm, which of the four orientations was more frequent within a given display. Three spatial conditions were used. In condition 1, the different orientations were randomly distributed over the surface. In the two other conditions, nine elements of the same orientation formed a group within one region of the texture. In condition 2, these grouped elements had the over-represented orientation. In condition 3, they had one of the three under-represented orientations. Grouping of the oriented elements has a facilitating effect when it contains the over-represented orientation, and a negative effect when it contains one of the under-represented orientations. There are good reasons to think that these interferences are decisional rather than sensorial.