bitmaps vs layers

What are the pros vs. cons of using bitmaps vs. layers. I've tried both, and performance wise if there is a difference the profiler isn't revealing it or I'm not doing something right. Programming wise it seems a wash, so is there any other consideration that would make one preferable over the other? I think one thing is that bitmaps support anti-aliasing for draw functions (after a certain version) and layers do not? Is this correct? Is there another alternative that beats both of these?