Wolbachia is an alpha-proteobacterial symbiont widely distributed in arthropods. Since the identification of Wolbachia in certain animal-parasitic nematodes (the Onchocercidae or filariae), the relationship between arthropod and nematode Wolbachia has attracted great interest. The obligate symbiosis in filariae, which renders infected species susceptible to antibiotic chemotherapy, was held to be distinct from the Wolbachia-arthropod relationship, typified by reproductive parasitism. While co-evolutionary signatures in Wolbachia-arthropod symbioses are generally weak, reflecting horizontal transmission events, strict co-evolution between filariae and Wolbachia has been reported previously. However, the absence of close outgroups for phylogenetic studies prevented the determination of which host group originally acquired Wolbachia. Here, we present the largest co-phylogenetic analysis of Wolbachia in filariae performed to date including: (i) a screening and an updated phylogeny of Wolbachia; (ii) a co-phylogenetic analysis; and (iii) a hypothesis on the acquisition of Wolbachia infection. First, our results show a general overestimation of Wolbachia occurrence and support the hypothesis of an ancestral absence of infection in the nematode phylum. The accuracy of supergroup J is also underlined. Second, although a global pattern of coevolution remains, the signal is derived predominantly from filarial clades associated with Wolbachia in supergroups C and J. In other filarial clades, harbouring Wolbachia supergroups D and F, horizontal acquisitions and secondary losses are common. Finally, our results suggest that supergroup C is the basal Wolbachia clade within the Ecdysozoa. This hypothesis on the origin of Wolbachia would change drastically our understanding of Wolbachia evolution.