The first part of this paper describes the quantitative characteristics of breeding habitats of sylviid warbler species, chiefly Sylvia and Phylloscopus, in Yorkshire, northern England, and in southern Sweden. The second part of the analysis is made up of the factors that permit or preclude the coexistence of various combinations of warbler species–vegetation structure, food density, and competitive regime. Vegetation structure was measured in 346 warbler territories in the 2 locations, including 151 Phylloscopus territories (P. trochilus, collybita and sibilatrix), 171 Sylvia territories (including S. communis, curruca, atricapilla, borin, nisoria and undata and 16 Hippolais icterina territories. From data on the distribution of the vegetation density over height in territories, discriminant function analysis generates new canonical habitat variables U1 and U2 that are independent of each other and reflect the correlation structure of the original habitat variables. Now each species' habitat niche can be shown as a family of "confidence ellipses" in the habitat plane of U1 and U2, and both species habitat preference and interspecific overlaps easily measured, within and between genera and within and between countries. As a result of this analysis, the following conclusions are drawn. The 3 English canopy—foraging Phylloscopus show some interspecific habitat segregation but more noticeably a great deal of overlap in preferred habitat. This overlap is reflected by the accuracy with which the specific occupant of a habitat patch can be predicted on the basis of the vegetation—structural characteristics of the patch: on average 2/3 of such predictions are verified, 1/3 is wrong. Similarly in Sweden, where 1 Phylloscopus drops out and is replaced by an additional sylviid, Hippolais icterina, predictions on specific habitat occupancy are correct just over 2/3 (72%) of the time. But the positions of habitat niches are rearranged between the 2 countries, because in Sweden the habitats of the absent collybita are taken over partially by sibilatrix but mostly by trochilus, and the newly added icterina occupies habitat that in England would be most typical of trochilus. Despite these differences, canopy warblers in both countries use very similar vegetational criteria in habitat selection, and predictions as to the specific occupant of a habitat based on discriminant function analysis in 1 country but tested by warbler data from the other country are nearly as successful as are within—country tests. Five Sylvia species were studied in each country, with 4 species in common. Interspecific habitat segregation is poorer than in Phylloscopus (54% of habitat occupants correctly classifiable). Four species occupy similar habitats in both countries with but minor habitat shifts, and in addition the Swedish nisoria occupies habitats that are structurally more similar to the English undata than to any other Swedish or English warblers. Again, cross—country predictions are just as successful, or as unsuccessful (@50% correct ), as within—country predictions. The second part of the paper describes and analyses the precise disposition of warbler territories within limited number of study sites in a range of habitat from low scrub to tall woodland. Pilot studies in English habitats show that 3 pairs of species: Phylloscopus trochilus and P. collybita, Sylvia atricapilla and S. borin, and Sylvia communis and S. curruca, species pairs with considerable overlap in preferred habitat type, are strongly interactive and show partial to complete interspecific territoriality. More detailed studies were conducted at 3 Swedish locations. In low woodland/tall scrub at Horby, 4 Sylvia species nested. Sylvia communis is shown to leave some potential territories in the field unoccupied, while S. borin occupies practically all habitat that ranks as highly acceptable. However, just 1 pair of atricapilla is present despite the fact that at least 4 additional atricapilla territories might be accommodated in suitable habitat, and likewise curruca fails to occupy all suitable habitat at the site. The presence of other Sylvia species, in particular the dominant borin, accounts for low densities of curruca and atricapilla, and a combination of statistical analysis of the disposition of territories and direct observation of supplanting and replacement through aggressive behavior confirms that interspecific interactions modify and qualify habitat acceptability. In tall woodland at Hoganas borin and atricapilla interact strongly over habitat, and atricapilla attains high occupancy levels only in habitat that is marginally acceptable to borin. The interaction between them is not affected by the presence or absence of Phylloscopus trochilus. At Bejershamn in low scrubland 5 Sylvia species and 1 each of Hoylloscopus and Hippolais breed. The species display a variety of levels of interspecific interaction, from a direct interspecific territoriality (borin—nisoria, communis—nisoria) to more subtle interactions of partial exclusion (borin—curruca). Here diffuse competition is evident (e.g., icterina is not found with borin and curruca together even though the last 2 occur frequently together and even though Hippolais icterina occurs with either separately). With increasing vegetation height in icterina habitat where more warblers coexist with overlapping territories, from 1— to 2— and 3—species combinations. But only 2/7 one—species combinations, 3/21 two—species combinations and 2/35 3—species combinations are common, stable and compatible. The remainder are precluded by differential habitat selection, by simple competition mitigated by direct behavioral interactions, or by diffuse competition with more subtle avoidance behavior. The role of song convergence in the maintenance of the interspecific interactions and the role of food density and distribution in the formation of compatible species combinations is documented, and further questions about the maintenance and evolution of the system are identified and discussed.