I've left it running now for about 12 hours on my machine, it reports: publish.request.4 has stopped core.track.4 has stopped So only two listeners have stoppped (same as after 1 hour) If I look at the queues in jmx-console, I have: publish.request: 527 messages core.track: 540 messages core.internal.publication: 11113 messages This makes sense, since core.internal.publications listeners are the slow ones - this seems to be pretty much a steady state. With these number of messages in publish.request and core.track It should easily be possible for one listener to get starved on the slow queues, since messages are prefetched in chunks (default chunk size of 150) to consumers. So I am happy with what I am seeing. Your code base is different to mine though and might account for the difference in behaviour. Right now the routing policy is not pluggable but there is a JIRA tasks for it. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3962721#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...