Dispersal behaviour and edge effects are two potential factors determining population densities, and both effects are likely to vary with patch size. However, the relative importance of these two effects may be hard to separate because they may produce similar patterns. Here, we separate these two effects on population densities of seven groups of arthropods on small islands. To separate dispersal behaviour and edge effects, we use the fact that the slope of the density–area relationships (DAR-slope) should change with the absolute rates of dispersal, as would occur in response to island isolation, whereas the edge effect is expected not to be dependent on island isolation. For lycosid spiders, parasitic wasps and possibly herbivorous Homoptera DAR-slopes changed between isolated and non-isolated islands, suggesting dispersal behaviour to be relatively more important for explaining variation in their densities. Other arthropods (ants and Collembola), typically those with a predicted low dispersal among islands, showed similar DAR-slopes between isolated and non-isolated islands consistent with dominant edge effects. For two groups (web spiders and Diptera) the results were inconclusive. We conclude that both migratory processes and edge effects should be considered in the evaluation of patch size and isolation on density–area relationships.