Abstract
The evolutionary processes responsible for adaptation and speciation on islands differ in several ways from those on the mainland. Most attention has been given to the random genetic drift that arises when a population is founded from just a few colonizing genomes. Theoretical obstacles to `founder effect speciation' are discussed, together with recent proposals for avoiding them. It is argued that although certain kinds of epistasis can facilitate the evolution of strong reproductive isolation, this favours divergence by selection as much as by random drift.