To say that your question is vague is an understatement.
I’m assuming that you are referring to a stable OSPF environment, but are NOT referring to the everyday glitches that happen to most networks from time to time.
Given the condition of a stable, working OSPF routed network, the most likely problem would be:
1) Lack of design controls (neighbors, Hello configurations, passwords, etc.) that would allow a piece of rogue equipment to interfere with an established routing environment.
2) Sections of the network that are routed by other protocols which “back-feed” the routing learned from OSPF into other protocols (RIP, RIP v.II, IGRP, EIGRP, BGP), and then re-injected into the network at a different point via OSPF.
3) Statically assigned values or weights that worked well with the entire network in place, but which break badly when part of the network goes down.
The real beauty of OSPF is that it’s truly vendor independent. I’ve implemented networks with equipment from Cisco, Lucent, 3Com, Nortel and others, and it works beautifully.
I hope that helps,