Several genomes of mycoplasmas have been sequenced and here we tried to retrieve the evolutionary relationships of nine species using a phylogenomic approach. Several methods were used to build phylogenetic trees based on protein sequence information, gene-order, and gene-content. We also utilized datasets composed of individual and concatenated sets of orthologous proteins, as well as with reduced unreliable alignment regions. Most of our results converge to a single topology, except for the trees built with both the maximum parsimony method and with the gene-order dataset. The gene-content dataset presented trees consistent with most nodes of the convergent tree, but in the gene-order dataset most internal branches were clearly saturated and unreliable. The topological difference between the trees obtained by the diverse methods could not be explained by regions with unreliable alignments or attributed to horizontal gene transfer among the genomes. It is possible that the incongruence between the methods could be associated with their differential sensibility in relation to certain evolutionary factors. Further analysis using other empirical genomic datasets would be necessary in order to better understand the basis of such conflicts.