In theory, one factor determining the rate and nature of the assembly of island biotas is the presence or absence of stepping stone islands, yet no field studies have demonstrated stepping stone function in practice. Krakatau, in Sunda Strait, is about equidistant from Java and Sumatra. Sebesi lies about half way between Krakatau and Sumatra, but no island intervenes between Krakatau and the nearest coast of Java. We assess the evidence that Sebesi has acted as an important stepping stone for Krakatau's recolonization since the devastating 1883 volcanic eruption. About a quarter of Krakatau's resident land birds, two-fifths of its reptiles, bats and land molluscs, and about two-thirds of its termites, pteridophytes, butterflies and spermatophytes are unknown on Sebesi, evidently having colonized without stepping stone involvement. Identifiable Sumatran taxa do not outnumber identifiable Javan ones on Krakatau, nor do historical distribution records indicate movement from Sebesi to Krakatau in animal groups. Krakatau's biota is not a subset of Sebesi's in predominantly anemochorous or thallassochorous plant groups, butterflies, reptiles or bats, and is only marginally so in termites. It is a subset in predominantly zoochorous spermatophyte groups, except Ficus species, and in birds and land molluscs. Comparison with a weaker stepping stone candidate, Panaitan, provides no evidence for a stepping stone role for Sebesi in butterflies or termites. We discuss the dispersal and establishment constraints on colonization by the groups involved, and conclude that, overall, Sebesi had little impact as a stepping stone. Instead, it is more probable that divergence of the environments of the two islands has led to an increasingly independent recolonization of Krakatau. © 2002 The Linnean Society of London, Biological Journal of the Linnean Society, 2002, 77, 275–317.