Abstract Aim Wallacea, the vast oceanic boundary between the Oriental and Australian regions, contains 122 true nectarivorous bird species. It is the contact zone of the Oriental sunbirds and the Australian honeyeaters, and at least three more true nectarivorous families are resident in the region. An island–bird matrix for Wallacea was tested for the presence of two patterns, nestedness and modularity. If the matrix is modular, it consists of a number of densely linked subgroups or modules of islands and birds, which are weakly interconnected. These modules are used as a new tool in analysing biogeographical boundaries in Wallacea. Location Wallacea, Indonesia. Methods We constructed an island–bird matrix for Wallacea and used two algorithms,aninhadoandsa, to test it for nestedness and modularity, respectively.aninhadocalculates the matrix temperature and provides a null model, andsais a module-detecting program based on simulated annealing. The results of thesawere compared with those from a hierarchical cluster analysis. Results The matrix had a nested pattern, as is commonly the case for island–species matrices. The SA detected four modules in Wallacea, each consisting of a group of islands sharing a group of nectarivorous birds. Thesaalgorithm produced a more detailed pattern of the area than did the hierarchical cluster analysis. Main conclusions Modularity and nestedness do not preclude each other as biogeographical patterns. The boundaries of the modules detected bysacompared well with major boundaries from the existing literature and showed a clear division of Wallacea into modules of birds and islands closely linked together. Thus modules are biogeographical units of islands sharing a specific nectarivorous fauna. For some research questions, we suggest that modules may be more appropriate biogeographical units than single islands or traditionally perceived archipelagos. The nectarivorous families showed distinctly different distributions, indicating variation in their colonization history and speciation processes. We recommendsaas a tool for detecting fine-grained biogeographical patterns.