Something to bear in mind is that display lists hold more than just vertex data. They can also hold serverside state changes and different kinds of geometry.
The very latest vertex array code using VBOs should be roughly the same in speed as a displaylist.
But the final word on the situation is that neither method of doing things
ever comes with a speed guarantee. In fact, nothing in OpenGL does. You need to choose the technique which best fits the problem you've got.
If you've got absolutely 100% static geometry then display lists are probably the simplest way to handle it.
Cas
