Islands are geographically defined as land masses completely surrounded by water, and island systems have been used as models for many biogeographic, ecological, and evolutionary theories ever since Darwin's pioneering efforts. However, their biological definition is complex. Over the past few decades these theories have been applied to many study systems that only share some geographic features with island systems. These features include spatial fragmentation, limited area, spatial and temporal isolation from adjacent parts of the system, and low connectivity between different parts within the system, to mention just a few. These systems vary in their form, the matrix that surrounds them, the factors defining their borders, the extent of insularity they impose on the different taxa, and their geological similarity to different types of actual islands. Here, I seek to understand whether such island‐like systems (ILS) function biologically as true islands. In the first part, I describe the wide diversity of ILS suggested in the literature and the variation in the features that define their insularity. In the second part, I review the extent to which the main theories of island biology are applicable to these systems: species–area and species–isolation relationships, community composition, evolutionary radiations, and the extent of endemism and genetic diversity. In the third and final part, I suggest a new conceptual framework within which to classify and study the biology of ILS, as well as practical future research directions. I conclude that the term ‘biological island’ is a multi‐faceted concept, loosely related to its geographical definition. As ILS are often less isolated than true islands, and their biological patterns are only partly similar to those of true islands (and even this is true only for some ILS) the use of the term ‘island’ to describe any isolated habitat is therefore inappropriate.