Context: Early-season immigration into arable fields by natural enemies is key for effective biocontrol, but little is known about the mechanisms underlying immigration processes. Objectives: Here we test the mass action hypothesis for ballooning spiders, stating that local immigration rates are positively related to the amount of spiders in the surrounding landscape. Methods: Immigration rates of spiders were assessed by sticky traps in remnant vegetation, in arable land 25–125 m from remnant vegetation, and in arable land further than 400 m from remnant vegetation. The experiment was conducted at 18 locations across two landscapes and repeated three times in a 2-week period in 2007 and 2008. Spider densities in crop and non-crop habitats were assessed by beat sheet sampling and used to calculate spider loads in landscape sectors around the experimental locations at five spatial scales. Results: Regression analysis indicated that immigration rates were influenced by meteorological variables and landscape context at 2 km and possibly beyond. Regression models that included spider load at relevant spatial scales received more statistical support from the data than models with the proportion of remnant vegetation and crops. Regression analysis further indicated that wheat and—to a lesser extent—remnant vegetation are important habitats for the recruitment of ballooning spiders. Conclusions: Our study provides support for the mass action hypothesis by showing that a combination of land-use variables with habitat specific spider densities allows the generation of functional cover types with greatly improved explanatory power.