While mass media are generally ascribed integrative potential, scholars assume negative effects on social cohesion for the Internet. Respective metaphors such as "filter bubbles" and "echo chambers" have meanwhile found their way into public discourse. However, empirical research that documents detrimental effects of digital fragmentation remains the exception. The state of research thus does not support fears about a stark loss of social cohesion due to the Internet. Yet there are groups on the fringe of the political spectrum that appear to be digitally fragmented. Given these heterogeneous results, the article systematizes theoretical assumptions and empirical findings and argues that it remains necessary to assess the extent and effects of digital fragmentation.