Hmmm. Murder decreases the 'expected utility' (cfg. life expentancy), so I think it would still be considered bad in some forms of consequentialism. The corner case - where expected utility would not change (much) would be e.g. shooting somebody who is falling off a cliff who will certainly not survive.
More general, it seems ethical systems are usually post-hoc organizing principles for our messy ethical intuitions. However, those intuitions are so messy, that for every simple set of rules, we can find some exception. Hence we get things like the trolley problem...
If it's worth saying, but not worth its own post, even in Discussion, it goes here.