The concept of modularity is fundamental to research in both evolutionary and developmental biology, though workers in each field use the idea in different ways. Although readily and intuitively recognized, modularity is difficult to define precisely. Most definitions of modularity are operational and implicit, particularly in developmental biology. Examination of several proposed definitions points to some general characteristics of developmental modules, for example their internal integration, and suggests the importance of devising a definition applicable at different levels of the biological hierarchy. Modules, like homologs, must be defined with respect to a specified level of the hierarchy, and a general definition should support both analyses of the evolving causal relationships between levels, and studies of the interconnections between modules of the same type. The designation of a developmental structure, process, or function as a “module” is a testable hypothesis; this hypothesis is confirmed in the case of the dorsal marginal zone of the amphibian gastrula, which acts as a morphogenetic module. Discussions of developmental modularity can provide a meeting place for developmental and evolutionary biologists by helping us articulate key questions at the intersection of the two fields, and design experiments to begin answering them.