I can't speak about every military in the world, but in general I think this attitude that military requirements are unconcerned about safety but are all about mission performance is completely invalid. I agree with what Nick has said, but the overall airworthiness situation goes beyond those specific examples.
For civilian and military airworthiness it is not about making something 100% safe. It is about making a product that carries an acceptable risk. To determine overall risk, you have to look beyond the failure modes and get into how the helicopter is being flown. The role itself comes with risks that airworthiness authorities must mitigate or accept. That becomes quite obvious in the military, such as the ballistics that has been discussed.
Another part of the role is what do you need your product to do to complete the role? A good example is the Bell 412. According to rumour, a civilian variant is limited to 5 degrees in a slope landing. It was determined that was insufficient for the Canadian military variant, so we had it increased to 10 degrees. The outcome of that is life of some parts is changed, so those parts don't go to civilian machines. Simple as that.
In Canada we've split airworthiness into three parts: operational, technical, and accident investigation. Apparently this model is being used by other countries and is gaining popularity. On the technical side its all about meeting specification requirements, proving equipment function, and determining safety. I spend much of my time reading through FAR and JAR documents in the process of obtaining military technical airworthiness.
A good example of how risk is mitigated is with OEI performance. For certain categories of civilian aircraft you have to load and/or fly your aircraft in such a way that you can lose an engine at anytime and successfully recover your aircraft. Other categories of civilian aircraft are certified to fly deep in the avoid curve with only one engine. The reason for the apparent disparity is because the mission needs to be done at a cost that is acceptable, and the risk when its just a pilot on board versus having many passengers is quite different.
In the military we are concerned with OEI performance. Whether you lose an aircraft to an RPG or an engine fail...you've still lost an aircraft, and perhaps the crew on board. We generate/validate data to help our pilots know what they can do when the engine fails. They and the commanders will know when the helicopter has poor OEI performance, and will consider the associated risks when mission planning. If it is fly into enemy fire to pick up some people who have no other out...probably not too concerned with OEI performance. If the mission is to demonstrate the capabilities of the machine to tens of thousands of people over/near populated areas...you'll probably find the orders demand a lighter helicopter.
Nick brought up a good point that is an unfortunate reality with todays budgets. The risk assessment now includes cost. A cheaper solution that comes with more risks may be accepted. Of course, this is not unique to military aviation, in fact it has been borrowed from civilian aviation, or more generally, from business.