Recent phylogenetic analyses of Saxifragales, based on DNA sequences of five plastid and nuclear ribosomal genes under parsimony and likelihood criteria, failed to resolve the branching order of the major lineages of the clade. This lack of resolution was attributed to the rapid radiation of Saxifragales. However, simultaneous origin of the major lineages was rejected by both parametric and non-parametric statistical tests, indicating that there is potential for resolving these relationships using additional analytical procedures and/or additional data. We reanalyzed the original data using a Bayesian approach, by implementing a Markov chain Monte Carlo process. Common-model and mixed-model analyses were con- ducted. Although Bayesian analyses did not fully resolve the phylogeny of Saxifragales, greater support was indicated for the relationships of Saxifragaceae, Crassulaceae, Haloragaceae, and Paeoniaceae. Relationships of ''lower hamamelid'' line- ages were inconsistently supported, including sister group relationships of Cercidiphyllum-Daphniphyllum and Altingiaiceae- Hamamelidaceae. Simultaneous origin of the major lineages was not supported. We compare parameter estimation in com- mon- and mixed-model Bayesian analyses and find striking differences that may affect the posterior probabilities of clades. Our application of Bayesian methods to discriminate simultaneous speciation from weak phylogenetic signal is a useful addition to the limited set of tools available for this purpose. We discuss important caveats for the interpretation of Bayesian posterior probabilities in this study, especially when a common model is assumed across data partitions.