Recent work indicates that feralisation is not a simple reversal of domestication, and therefore raises questions about the predictability of evolution across replicated feral populations. In the present study we compare genes and traits of two independently established feral populations of chickens (G. gallus) that inhabit archipelagos within the Pacific and Atlantic regions to test for evolutionary parallelism and/or divergence. We find that these two feral populations share close genetic similarities despite the lack of any current gene flow between them. Next, we used genome scans to contrast the targets of feralisation (selective sweeps) between the two independently feral populations from Bermuda and Hawaii. Three sweep loci (each identified by multiple detection methods) were shared between feral populations, and this overlap is inconsistent with a null model in which selection targets are randomly distributed throughout the genome. In the case of the Bermudian population, many of the genes present within the selective sweeps were either not annotated or of unknown function. Of the nine genes that were identifiable, five were related to behaviour, with the remaining genes involved in bone metabolism, eye development, and the immune system. Our findings suggest that a subset of feralisation loci (i.e. genomic targets of recent selection in feral populations) are shared across independently-established populations, raising the possibility that feralisation involves some degree of parallelism or convergence. A clearer understanding of whether these reflect selection for similar functional traits (‘feralisation syndromes’) will require elucidating genotype-phenotype relationships in any populations being compared.