MrBayes / News: Recent posts

MrBayes 3.2.4 Released

Dear all,

We are happy to announce a new release of MrBayes: version 3.2.4.
This release contains critical bug fixes. It is highly recommended for all users to update.
You can download it from

Release note:
1. Correct the proposal ratio in the node slider move for non-clock tree.
2. Fix reporting ancestral states under GTR+I+G model.
3. Fix the sliding window in the move of updating omega in codon model.
4. Fix the likelihood and proposal ratio of NaN in the multi-species coalescent analysis.
5. Correct screen output messages for a few commands.
6. Fix the interaction of MrBayes with Beagle for the Windows executable.
7. Fix prior pdf of relative rates (rate multipliers).... read more

Posted by Chi Zhang 2014-11-17

MrBayes 3.1.2 Released

We are very happy to announce the 3.1.2 release of MrBayes. It can be characterized as a maintenance release. We have looked at all bug reports submitted to according to the instructions on the mrbayes web site ( and we have fixed all problems we have been able to replicate. Among the more important bug fixes and changes:

1. Constrained partitions previously could be broken by the LOCAL move under some conditions. If you have run constrained analyses or pulled ancestral state reconstructions for a constrained node using previous versions of MrBayes, please make sure (using sumt for instance) that all samples contain the constrained node(s). If not, you should rerun these analyses, or filter out the typically few samples that violate the constraint(s). This bug has been fixed in 3.1.2.... read more

Posted by Paul van der Mark 2005-12-23

MrBayes 3.1 Released

We are pleased to announce the release of version 3.1 of MrBayes, the first non-beta version of MrBayes 3.

In version 3.1, all bugs known to us have been addressed and we have made numerous minor improvements. Several major additions have also been made. Perhaps most important of these, MrBayes can now run several independent analyses simultaneously and use the samples from these to calculate convergence diagnostics on the
fly. This makes it possible, for instance, to stop a run automatically when convergence has (apparently) been reached. The "sumt" and "sump" commands have been
augmented to compute summary results from several independent runs, including convergence diagnostics for all model parameters. Other major additions include the possibility to enforce topological constraints and to infer site rates or ancestral states for particular nodes in the tree. On popular demand, we have implemented the GTR model for protein data. Among other things, this allows the user to specify any fixed reversible rate matrix for protein data using the "prset statefreqpr=fixed(...) aarevmatpr=fixed(...)" mechanism.

Posted by Paul van der Mark 2005-05-17