Most empirical evidence suggests that balancing selection does not counter the effects of genetic drift in shaping post-bottleneck MHC genetic diversity when population declines are severe or prolonged. However, few studies have been able to include data from historical specimens, or to compare populations/species with different bottleneck histories. In this study we examined MHC class II B and microsatellite diversity in four New Zealand passerine (songbird) species that experienced moderate to very severe declines. We compared diversity from historical samples (collected circa 1884-1938) to present day populations. Using a Bayesian framework, we found that the change in genetic diversity from historical to contemporary samples was affected by three main factors: 1) whether the data were based on MHC or microsatellite markers, 2) species (as a surrogate for bottleneck severity) and 3) whether the comparison between historical and contemporary samples was made using historical samples originating from the mainland or using historical samples originating from islands. The greatest losses in genetic diversity occurred for the most severely bottlenecked species, particularly between historical mainland and contemporary samples. Additionally, where loss of diversity occurred, the change was greater for MHC genes compared to microsatellite loci.