This paper describes the route, speed and mode of colonisation of oaks by integrating a number of independent analyses using molecular ecology, palaeoecology and simulation modelling approaches. Using a synthetic map of the contemporary distribution of chloroplast DNA (integrating several published and unpublished data sets and describing variation in 1468 trees from 313 autochthonous stands of Q. robur and Q. petraea from Britain and Ireland), and considering the postglacial topographic landscape, the most likely routes of postglacial colonisation across the British Isles are suggested. The overall pattern of these directions agrees with previous interpretations, but several routes, particularly those into Ireland, differ from previous interpretations and benefit here from using a single synthesised data set. Interestingly, the Atlantic oakwoods appear to have been colonised by individuals bearing a single haplotype (type 12). Two palaeoecology data sets, published separately for Britain and Ireland, are synthesised here and used to infer the timing of first arrival of oaks across the British Isles (between 9500 and 6000 years before present). The maximum observed colonisation speed within the British Isles is approximately 500 m year(-1) in central and southern England. Outputs from a simulation model, which mimics postglacial colonisation processes, and which has been parameterised for the colonisation rate observed from the pollen core record and contemporary cpDNA structure, predict that the rapid colonisation rate observed, for at least the southern portion of the British Isles, can only be achieved via very rare (an approximate frequency 0.01%), very long distance seed dispersal events (up to 100 km). Potential agents of such dispersal events are birds or major meteorological disturbances, e.g. hurricanes. Additional simulation modelling and genetic analysis of latitudinally stratified populations indicate that non-synchronous colonisation fronts, topographic barriers and temperature related survival may also have had an effect on the speed of migration and resulting genetic structure. Finally, in an attempt to record predicted long distance seed dispersal events, a novel curve fitting technique is applied to molecular parentage assignment data for field established seedlings from a contemporary population. A notable discrepancy is recorded between contemporary field estimates (just over 1 km) and those predicted by simulation modelling, and is discussed in detail. A concluding section describes future research priorities.< Réduire