Spiders are an abundant and diverse group of generalist predators in arable fields. Knowledge on what landscape and site factors affect this group can be valuable for efforts to reduce biodiversity loss in agricultural landscapes and can have implications for natural pest control. We investigated the impact of landscape and site factors on epigeic spiders in 29 winter oilseed rape fields (Brassica napus, OSR) embedded in differently structured landscapes in an agricultural region east of Vienna (Austria). Landscape factors included proportions of non-crop areas, woody areas and fallows, lengths of road-side strips and hedges, and landscape diversity at different spatial scales (r=250–2000 m). Site factors included OSR stand density, soil index, soil cultivation intensity, nitrogen fertilisation level, OSR vegetation cover in late autumn, and insecticide applications. Data were analysed using regression, ordination, and variation partitioning. Different characteristics of spider assemblages responded to different landscape factors at different spatial scales. Observed species richness showed the strongest positive reponse to proportions of woody areas at rather small scale (radius 500 m), but the relation remained significant up to the 1250 m radius. Standardised species richness was positively related to non-crop area at the smallest scale (radius 250 m). Activity density was positively related to length of road-side strips with maximum effects at large scale (radius 1750 m) and non-crop area (radius 750 m). Site characteristics (stand density, insecticide applications, and late autumn ground cover) and landscape factors (woody areas and fallows at radius 500 m) were similarly important for explaining species composition. We interpret the scale-dependency of relations as the result of differences in dispersal power of the studied spider species. These results demonstrate the important, scale dependent influence of natural and semi-natural habitats on spider assemblages in arable fields.