In patchily distributed species dispersal connects local populations into metapopulations. Reliable quantifications of dispersal are therefore crucial to understanding the population dynamics and genetic structure of such metapopulation systems. The great reed warbler (Acrocephalus arundinaceus) inhabits eutrophic lakes and has a patchy breeding distribution. In this study we investigated the dispersal pattern of the great reed warbler based on an extensive capture-recapture effort covering a large census area (22,500 km(2)). At two adjacent breeding sites (10 km apart) in southern Central Sweden, the "main study area", we ringed the majority of adult and nestling great reed warblers between 1902 and 1999. In 1998 and 1999, we opportunistically searched for territorial males at the majority of the Swedish breeding sites, and were able to examine about 56% of all males in the region. Analyses of recaptured males demonstrated that philopatry predominated. Sixty-nine percent of the recruiting nestlings returned to breed in the main study area (their natal area), and 92% of the resighted adults were found at the same breeding locality in both study years. Breeding dispersal was significantly more restricted than natal dispersal. Additional data on natal and breeding dispersal within the main study area in 1992-1999 suggested that females were as philopatric as males. The overall high level of philopatry, with only occasional longer dispersal distances documented, yielded a root-mean-square dispersal distance of 33 km per generation. High philopatry, short dispersal distances and similar dispersal patterns of male and female great reed warblers contrast the findings among birds in general, but conform to data of species having patchy breeding habitat and isolated populations. Restricted dispersal suggests limited gene flow even among several Swedish populations, which is in line with some previous findings of the population ecology of the great reed warbler: (1) structured mtDNA lineages among European populations (2) small-scale population differences in song patterns; and (3) low genetic variation and occurrence of inbreeding depression in our main study population.