We have an EIGRP network with about 300 layer three devices (3750, 6500, 3845, 7604) and are starting a migration to OSPF project.
One of the big questions is whether we want to have more than Area 0. A big reason to carve up the network into areas would be to have summarization, but that's not going to happen in our network because we've got a class-B /25 subnets scattered all over the place and will not be readressing due to extreme labor in all aspects of the idea.
So since summarization is not a reason to have multiple areas, I'm of a mind to just have one area for our OSPF environment. Other network engineers are warning against a single area, some saying there should only be 50 L3 devices in an area. My idea is to keep is simple, and I don't think with the CPU power of these devices, that having all our routes in one area is a problem.
I found the attached document quite interesting concerning areas.