The mosquito Anopheles gambiae s.s. is a primary malaria vector throughout sub-Saharan Africa including the islands of the Comoros archipelago (Anjouan, Grande Comore, Mayotte and Mohéli). These islands are located at the northern end of the Mozambique Channel in eastern Africa. Previous studies have shown a relatively high degree of genetic isolation between the Comoros islands and mainland populations of A. gambiae, but the origin of the island populations remains unclear. Here, we analyzed phylogenetic relationships among island and mainland populations using complete mitochondrial genome sequences of individual A. gambiae specimens. This work augments earlier studies based on analysis of the nuclear genome. We investigated the source population of A. gambiae for each island, estimated the number of introductions, when they occurred and explored evidence for contemporary gene flow between island and mainland populations. These studies are relevant to understanding historical patterns in the dispersal of this important malaria vector and provide information critical to assessing their potential for the exploration of genetic-based vector control methods to eliminate this disease. Phylogenetic analysis and haplotype networks were constructed from mitogenome sequences of 258 A. gambiae from the four islands. In addition, 112 individuals from seven countries across sub-Saharan Africa and Madagascar were included to identify potential source populations. Our results suggest that introduction events of A. gambiae into the Comoros archipelago were rare and recent events and support earlier claims that gene flow between the mainland and these islands is limited. This study is concordant with earlier work suggesting the suitability of these oceanic islands as appropriate sites for conducting field trial releases of genetically engineered mosquitoes (GEMs).