The ability to construct recombinant poxviruses has been central to deciphering host responses to viral infection and to the clinical investigation of viral-vectored vaccines. Methods to develop recombinant poxviruses have been well established since the first recombinant poxviruses were generated in the early 1980’s and rely on homologous recombination between exogenous DNA and viral genomes within infected cells. The length of the DNA sequence shared by the virus and transfer constructs is known to affect the rate of homologous recombination, but few other factors have been explored. Our research sought to generate recombinant VACVs expressing a foreign gene from A53R, a non-functional gene. However, using our initial construct designed to target the A53R locus, we were unable to integrate a linear fragment of DNA into the viral genome, a process requiring simultaneous recombination events between the virus and the two homology arms flanking the foreign gene. Instead, we were only able to obtain recombinants that resulted from integration of circular plasmids. Analysis of these viruses suggested that recombination was only occurring efficiently at one of the two homology arms (both of which were relatively short at 250bp), with the arm of lowest G+C (23%) seemingly disfavoured. To test if G+C, length, or both were the limiting factor/s two new transfer constructs were made targeting A53R, one with the original homology arms extended to included regions of higher G+C and another where the arm with lowest G+C was moved to improve G+C but keeping the length at ~250bp. We were able to generate the desired virus with both of these new constructs on the first attempt. Therefore low G+C content in homology arms represents a previously undocumented factor that can limit the success of making recombinant VACV.