It is suspected that this is the root cause of a second NPE that can be observed in this way: Run the ''switchyard-rules-interview' quickstart under concurrent use (10 threads for 10 requests each reliably produces an NPE on my machine). Note that this test will raise an NPE in the OutboundHandler. Closer inspection reveals that this NPE is the result of handling an earlier thrown NPE, thought to be the KIE session access problem.