Extant clades may differ greatly in their species richness, suggesting differential rates of species diversification. Based on phylogenetic trees, it is possible to identify potential correlates of such differences. Here, we examine species diversification in a clade of 82 tropical African forest butterfly species (Cymothoe), together with its monotypic sister genus Harma. Our aim was to test whether the diversification of the Harma–Cymothoe clade correlates with end-Miocene global cooling and desiccation, or with Pleistocene habitat range oscillations, both postulated to have led to habitat fragmentation. We first generated a species-level phylogenetic tree for Harma and Cymothoe, calibrated within an absolute time scale, and then identified temporal and phylogenetic shifts in species diversification. Finally, we assessed correlations between species diversification and reconstructed global temperatures. Results show that, after the divergence of Harma and Cymothoe in the Miocene (15 Mya), net species diversification was low during the first 7 Myr. Coinciding with the onset of diversification of Cymothoe around 7.5 Mya, there was a sharp and significant increase in diversification rate, suggesting a rapid radiation, and correlating with a reconstructed period of global cooling and desiccation in the late Miocene, rather than with Pleistocene oscillations. Our estimated age of 4 Myr for a clade of montane species corresponds well with the uplift of the Eastern Arc Mountains where they occur. We conclude that forest fragmentation caused by changing climate in the late Miocene as well as the Eastern Arc Mountain uplift are both likely to have promoted species diversification in the Harma–Cymothoe clade. Cymothoe colonized Madagascar much later than most other insect lineages and, consequently, had less time available for diversification on the island. We consider the diversification of Cymothoe to be a special case compared with other butterfly clades studied so far, both in terms of its abrupt diversification rate increase and its recent occurrence (7 Myr). It is clear that larval host plant shift(s) cannot explain the difference in diversification between Cymothoe and Harma; however, such a shift(s) may have triggered differential diversification rates within Cymothoe. © 2013 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, ●●, ●●–●●.