Sorry, I don't have a good answer for you. :( The RMI-stub caching issue can perhaps be fixed in Remoting using an approach similar to what I did in JNDI. I can only speculate on how Remoting could handle the tx timeout scenario you describe -- perhaps catch the InterruptedException, check the status of any tx associated with the thread and throw a meaningful exception on to the client? That's just speculation though, and seems hacky -- the Remoting guys would need to give you a better answer.