Summary1. An explicitly spatial sampling approach was employed to test the null hypothesis that the predation on slugs by the carabid beetle Pterostichus melanarius (Illiger) was opportunistic. 2. The beetles and slugs were sampled across a nested series of grids of sampling points, in a ®eld of winter wheat during June and July 1997. 3. The spatial distribution of all slugs in June was found to change with the scale of the sampling grid, from random on the 0.25 m scale, through aggregation at 1 m, to random at 4 m. At the highest scale of 16 m, the slugs were signi®cantly spatially aggregated. 4. The distribution of beetles in June was also spatially dynamic, with randomness observed at the 4 m and 8 m scales. At 16 m, signi®cant aggregation was observed. 5. The dynamic distributions of slugs and beetles, at 16 m, were found not to be associated with, and thus were not determined by, soil or crop factors. 6. Comparison of slug and beetle populations showed, however, that the distributions at 16 m were dynamically associated with each other. In June where there were many slugs there were also many carabids, whilst in July where there were many carabids there were few slugs. 7. Approximately 11% of the beetles sampled across the 16 m grid in June and July were found to have ingested slug protein, following intensive enzyme-linked immunosorbent assay (ELISA) testing. 8. The spatial distribution of these slug-positive beetles was signi®cantly associated with the distribution of the larger slug classes, over 25 mg. Where there were many large slugs in June there were many slug-positive beetles. Conversely, in July few large slugs were found where there were many slug-positive beetles. 9. Parametric analysis revealed that these changes in the large slug class, at each sampling point between June and July (growth), were negatively related to the local numbers of slug-positive beetles, and that growth declined as the local numbers of beetles increased. 10. These ®ndings suggest that predation was not opportunistic, but direct and dynamic, falsifying the null hypothesis. Moreover, this predation elicited signi®cant changes in the spatial distribution and local density of the slugs, in a manner that may be termed spatially density dependent.