Huge single mesh or many little ones?

I don’t know if I should make the gamelevels as a single huge mesh with, say, 50.000 triangles or make many parts of it, with something like 500 triangles each, and then build them together in Unity. I know that with the latter option I could reuse the parts to make other levels, but say I want to have just one level and I care only about performance. So what is the best option for a good performance?

Sorry for bad english.

For performance, it’s definitely better to have separate little parts. That way the camera can only render the parts where it’s looking (say only about 1000 polygons) and doesn’t have to render the entire 50000 polygons.
This is also really important for other things like occlusion culling which hugely improves performance.

But you don’t necessarily have to build the level in unity, you can make the level in your 3d modelling app also. Just make sure the parts are separate submeshes and they show up as individual child gameobjects in unity!