The difference is that the workitem is set to state aborted instead of completed. And you could for example access the work item state in an on-exit action on the work item and do something in that case (throw an exception, set a process variable). It is correct however that the process continues, regardless of whether you completed or aborted the work item. This is actually something that we have foreseen as a future extension but is not yet implemented in the editor: I would like to make it possible to have a separate outgoing connection for when the work item has been completed. This would then allow you to specify a path that needs to be followed when the work item has been aborted. But we need support for that in the editor before being able to do that. So for now, I would suggest sticking to complete, and simply pass any information using the result parameters, either normal completion information or error information. You can then map this info to process variables and use anytime further in the process. Kris Quoting jschmied <nabble(a)juergenschmied.de&gt;: http://drools-java-rules-engine.46999.n3.nabble.com/WorkItem-meaning-of-a... Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm