Considerable conflict remains in the literature as to the position of the root of placental mammals, and the placement of several intra-ordinal groups. Debate continues over the use of DNA or amino acids datasets and over the use of Supertree or Supermatrix approaches. Known phenomena exist within mammal data that complicate the reconstruction of phylogeny. These include (but are not limited to), variation in longevity, body size, metabolic rates, and germ-line generation time that result in variation in mutation rates and composition biases. Previous attempts to resolve the placental mammal phylogeny have used homogeneous evolutionary models that cannot capture and adequately describe these features across the species sampled. In this thesis I explore the properties of different datasets and data types and their suitability to the resolution of the mammal phylogeny at different depths: (i) the position of the root of the placental mammals, and (ii), the intraordinal placements within the Laurasiatheria. The datasets tested were (i) mitochondrial and nuclear data types, (ii) previously published datasets for mammals, and (iii), datasets I assembled specifically for analyses at different phylogenetic depths. I propose and apply the use of heterogeneous models to resolve the position of the root of the placental mammal phylogeny to these datasets. Reconstruction of a robust mammal phylogeny provides us with an essential framework for understanding the molecular underpinnings of adaptation to environment. The placental mammals display a huge variations in life traits such longevity, body size and DNA repair efficiency, since they emerged ~100 million years ago. With this robust phylogeny, I set out to determine the level of adaptive and non-adaptive processes acting on a set of mammal genes that are linked with longevity and cancer. The results of these analyses yield important insights into data and model suitability, and provide strong evidence for a single hypothesis for the rooting of placental mammals. These results also show that Laurasiatheria intra-ordinal placements are not fully resolved and additional sampling from this diverse clade is required. Using this resolved phylogeny, specific molecular adaptations and non-adaptive mechanisms were identified in the mammalia for a set of telomere-associated genes.