I'm not entirely sure whats going on here but I suspect something along these lines: Request 1 All of the context variables/bean variables are empty so the Factory method is called for each. Request 2 All of the context variables are empty, all of the bean variables are initialised (to the values from Request 1) as the bean is in the session scope. As the first context variable is null the Factory method is called for it and the bean variable updated. Bean values which are not null are outjected from the bean - thus all three context variables are now not empty. With the first one updated and the second two not. I think it would be fair to say its a 'non-standard' pattern to outject into a shorted-lived scope than the scope of the outjecting bean. I don't use the index strategy so can't really discuss patterns for it. If you want to use with PAGE scope (which is the best way to use I suppose) you need to use a SLSB to control it. Otherwise I would suggest using the STRING strategy with an unchanging unique identifier on the outjected object or the OBJECT strategy and use either the provided converters (for EJB3 entities, enums) or write your own. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3981823#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...