The rapid accumulation of sequenced genomes offers the chance to resolve longstanding questions about the evolutionary histories, or phylogenies, of groups of organisms. The relatively rare occurrence of large-scale evolutionary events in a whole genome, events such as genome rearrangements, duplications and losses, enables us to extract a strong and robust phylogenetic signal from whole-genome data. The work presented in this dissertation focuses on models and algorithms for whole-genome evolution and their use in phylogenetic inference. We designed algorithms to estimate pairwise genomic distances from large-scale genomic changes. We refined the evolutionary models on whole-genome evolution. We also made use of these results to provide fast and accurate methods for phylogenetic inference, that scales up, in both speed and accuracy, to modern high-resolution whole-genome data. We designed algorithms to estimate the true evolutionary distance between two genomes under genome rearrangements, and also under rearrangements, plus gains and losses. We refined the evolutionary model to be the first mathematical model to preserve the structural dichotomy in genomic organization between most prokaryotes and most eukaryotes. Those models and associated distance estimators provide a basis for studying facets of possible mechanisms of evolution through simulation and application to real genomes. Phylogenetic analyses from whole-genome data have been limited to small collections of genomes and low-resolution data; they have also lacked an effective assessment of robustness. We developed an approach that combines our distance estimator, any standard distance-based reconstruction algorithm, and a novel bootstrapping method based on resampling genomic adjacencies. The resulting tool overcomes a serious and long-standing impediment to the use of whole-genome data in phylogenetic inference and provides results comparable in accuracy and robustness to distance-based methods for sequence data. Maximum-likelihood approaches have been successfully applied to phylogenetic inferences for aligned sequences, but such applications remain primitive for whole-genome data. We developed a maximum-likelihood approach to phylogenetic analysis from whole-genome data. In combination with our bootstrap scheme, this new approach yields the first reliable phylogenetic tool for the analysis of whole-genome data at the level of syntenic blocks.