I have to comment of the last post. If it is truly not possible to teach students to adequately enter the auto after a true surprise engine failure, then how do you account for anyone ever getting into the autos you state are always survivable?
I have worked with dozens of students and low-time new hire pilots and all have been able to get an R22 or R44 into the auto following a surprise 'throttle chop' or they were not considered to be ready for the flight test/job.
It is true that carb ice and distractions from the aviating part of the job are large dangers for pilots of these types but that doesn't mean proper response isn't possible.
In regards to discussion further up the list, I have always performed autos to touchdown only on surfaces that have been inspected by myself and that I deem as minimal risk to the aircraft. Surprise practice engine failures off airfield and in areas I haven't surveyed are always taken to the hover power recovery to minimize risk. We also strived to avoid being predictable. Your student should NOT be able to anticipate when the engine will 'quit' on a training flight. These off-airfield autos were performed to teach just what was discussed above by Tain't Natural, a good entry to a stabilized auto that reaches a good landing spot. This is a survivable auto.
If I really thought that an engine failure was a guaranteed death sentence for these pilots then I would have a very hard time sleeping at night.
For myself, I had a student test my ability to enter an auto in an R22. While talking with my hands in the cockpit she had a small mental blunder and rolled off the throttle while in a max t/o power climb through a little over 1000'. As I heard the engine die I took control, entered the auto and was heading for a field. It wasn't until about halfway down that I noticed the engine idling (the horn covered the noise for a few seconds) and rolled up throttle to fly away. I was surprised but I was happy to learn that I would get the pole down.