We describe a general methodological framework for reconstructing ancestral genome segments from conserved syntenies in extant genomes. We show that this problem, from a computational point of view, is naturally related to physical mapping of chromosomes, and benefits from using combinatorial tools developed for this problem. We develop this framework into a new reconstruction method considering conserved gene clusters with similar gene content. We implement and apply it to datasets of mammalian genomes. Compared to other bioinformatics methods for ancestral genome segments reconstructions, this one is stable: it gives convergent results using several kinds of data and different levels of resolution; it is reliable: all predicted ancestral regions are well supported; and it gives results that come very close to cytogenetics studies. It suggests the principle we propose, although based on a bioinformatics approach, relies on fundaments that are close to the ones used to analyze cytogenetic data.