Background: Recent studies of selection on mitochondrial (mt) OXPHOS genes suggest adaptation due mainly to environmental variation. In this context, Tunisian hares that display several external phenotypes with phylogenetically rather homogenous gene pool and shallow population structure provide a good precondition to detect positive selection on mt genes related to environmental/climatic variation, specifically ambient temperature and precipitation. Results: We used codon-based methods along with population genetic data to test for positive selection on ATPase synthase 6 (ATP6) and NADH dehydrogenase 2 (ND2) of cape hares (Lepus capensis) collected along a steep ecological gradient in Tunisia. We found significantly higher differentiation at the ATP6 locus across Tunisia, with sub-humid Mediterranean, semi-arid, and arid Sahara climate than for fourteen unlinked supposedly neutrally evolving nuclear microsatellites and mt control region sequences. This suggested positive selection on ATP6 sequences, which was confirmed by several codon-based tests for one sequence site that together with a second site translated into four different amino acids. Positive selection on ND2 sequences was also confirmed by several codon-based tests. The corresponding frequencies of the two most prevalent variants at each locus varied significantly across climate regions, and our logistic general linear models of occurrence of those proteins indicated significant effects of mean annual temperature for ATP6 and mean minimum temperature of the coldest month of the year for ND2, independent of geographical location, annual precipitation, and the respective co-occurring protein at the second locus. Moreover, presence of the ancestral ATP6 protein, as inferred from phylogenetic networks, was positively affected by the simultaneous presence of the derived ND2 protein and vice versa, independent of temperature, precipitation, or geographic location. Finally, we obtained a significant coevolution signal for the ancestral ATP6 and derived ND2 sequences and vice versa. Conclusions: positive selection was strongly suggested by the population genetic approach and the codon-based tests in both mtDNA genes. Moreover, the two most prevalent proteins at the ATP6 locus were distributed at significantly varying frequencies across the study area with a significant effect of mean annual temperature on the occurrence of the ATP6 proteins independent of geographical coordinates and the co-occuring ND2 protein variant. For ND2, occurrence of the two most frequent protein variants was significantly influenced by the mean minimum temperature of the coldest month, independent of the co-occurring ATP6 protein variant and geographical coordinates. This strongly suggests direct involvement of ambient temperature in the adaptation of the studied mtOXPHOS genes.