We examine how spatial subdivision of predator–prey systems affects colonization processes in metapopulations. Dynamics of the herbivorous spider mite Tetranychus urticae (prey) and the predatory mite Phytoseiulus persimilis are highly unstable on isolated bean plants (Phaseolus lunatus) and ultimately result in extinction of prey and predators. Assembling a collection of 90 plants without any dispersal barriers (a super-island experiment) does not modify the persistence of the predator–prey system. Subdividing the system into a metapopulation with barriers for dispersal (a collection of eight islands with 10 plants per island) leads to persistence of the predator–prey dynamics for many generations. In this paper, we use the time series of colonization events and prey and predator densities from the super-island and metapopulation experiments to understand how colonization processes of prey and predatory mites are altered by spatial subdivision. Using survival analysis, we estimate how prey and predator colonization probability is affected by densities of the colonist pool at different distances from the target plant. Contrasting the results from the super-island and metapopulation experiments reveals that spatial subdivision affects the discovery rate of prey outbreaks by predatory mites and differentially affects colonization by prey and predators. Prey colonization is primarily determined by local densities of prey in spatially subdivided systems, whereas predator colonization retains primarily “global” influences. Our analysis of colonization processes suggests mechanisms accounting for stability in the metapopulation experiments and provides the quantitative basis for the development of colonization functions to explore these mechanisms in predator–prey models of acarine systems.