Several hypotheses have been advanced for the origin and diversification of the bat fauna of the New World. Traditional models considered one of the families (Vespertilionidae) to have had a North American origin, whilst the diversification of other seven families was thought to have occurred in South America. Present-day patterns of diversity are the result, according to these hypotheses, of the mixing of faunas, mostly coinciding with the Great American Biotic Interchange (GABI) of the Americas. Recent research has challenged the traditional model and has posited the possibility of dual centres of diversification (in South and North America) for families traditionally considered of strict South American origin. Here we examine the latitudinal patterns of bat diversity and endemism at the genus level and show that present-day patterns are more consistent with this dual-diversification hypothesis than with the traditional models. We document an asymmetrical latitudinal gradient of genus richness with a plateau near the equator, but with a decline in Central America and Mexico; however, in this area the relative number of genera, considering the species richness, is higher than expected by chance, as shown by a proper null model. The distribution and identity of endemic genera are consistent with models considering a North American origin for two of the families (Vespertilionidae and Natalidae), an early arrival to South America for another family (Molossidae), and a complex diversification process, involving two centres of diversification, for at least two of the families of the Noctilionoidea (Mormoopidae and Phyllostomidae) and possibly for the Emballonuridae. The recently identified Panamanian Realm (including southern Mexico and Central America) seems to have played a significant role in the diversification of New World bats.