The ability to obtain ancient DNA (aDNA) from extinct species has greatly improved our ability to study their evolutionary histories. Most aDNA studies focus on mitochondrial DNA for both population-level and species-level insights, due to its relatively high abundance and increased obtainability. Although several approaches are available to assemble mitogenomes, high levels of DNA damage and non-target contaminating DNA make many approaches unsuitable for aDNA. Further complicating matters for most extinct species is the lack of a sufficiently closely related extant relative to act as a reference. Iterative mapping can assemble mitogenomes without using a close reference sequence, and has been applied to aDNA from various extinct species. Despite its widespread use, the accuracy of the reconstructed assemblies is yet to be comprehensively assessed. Here, we investigated the influence of (i) mapping software (Burrows–Wheeler Alignment [BWA], MITObim and NOVOplasty), (ii) parameters and (iii) bait reference phylogenetic distance on assembly accuracy using two simulated datasets: spotted hyena and various mammalian bait references; southern cassowary and various avian bait references, and two empirical datasets: extinct cave hyena; two extinct sloth species. Specifically, we assessed the accuracy of results through three parameters: pairwise distance to a reference conspecific mitogenome, number of incorrectly inserted base pairs and total length of the assembly. We found large discrepancies in the accuracy of reconstructed assemblies using different mapping software, parameters and bait references. Results also varied depending on the dataset, thereby making broad, generalised conclusions difficult. However, the overall most accurate results were obtained with either BWA using a minimum mapping quality of 20 and relaxed mismatch parameters (-n 0.001 -o 2), or MITObim using a mismatch value of 3 or 5. We also show accuracy could be further improved by combining results from multiple bait references. Our study provides information on how divergent mitogenomes are best reconstructed from aDNA. By obtaining the most accurate reconstruction possible, one can be more confident as to the reliability of downstream analyses, and the evolutionary inferences made from them.