Phylogenetic trees of present-day species allow investigation of the rate of evolution that led to the present-day diversity. A recent analysis of the mammalian phylogeny challenged the view of explosive mammalian evolution after the Cretaceous-Tertiary (K/T) boundary (65 Mya). However, due to lack of appropriate methods, the diversification (speciation minus extinction) rates in the more recent past of mammalian evolution could not be determined. In this paper, I provide a method that reveals that the tempo of mammalian evolution did not change until ∼ 33 Mya. This constant period was followed by a peak of diversification rates between 33 and 30 Mya. Thereafter, diversification rates remained high and constant until 8.55 Mya. Diversification rates declined significantly at 8.55 and 3.35 Mya. Investigation of mammalian subgroups (marsupials, placentals, and the six largest placental subgroups) reveals that the diversification rate peak at 33-30 Mya is mainly driven by rodents, cetartiodactyla, and marsupials. The recent diversification rate decrease is significant for all analyzed subgroups but eulipotyphla, cetartiodactyla, and primates. My likelihood approach is not limited to mammalian evolution. It provides a robust framework to infer diversification rate changes and mass extinction events in phylogenies, reconstructed from, e.g., present-day species or virus data. In particular, the method is very robust toward noise and uncertainty in the phylogeny and can account for incomplete taxon sampling.