What do you mean by outbound any any rules? You are expecting to block only on the inbound?
You need to fully describe the architecture. Are both domains AD, or is one? What client OS's? Are you using WINS (if so, you need to track down the wins replication ports, tcp 42 IIRC).
To start, you will need 135-139,445 open for tcp and udp, for windows filesharing and auth to work. Depending of the state of win2k rollout, you might need to open up global catalog ports, ldap (tcp 389), kerberos (88tcp/udp), etc.
http://www.microsoft.com/technet/treeview/default.asp?url=/technet/prodtechnol/exchange/exchange2000/reskit/resguide/appendb.asp
For what it is worth, implementing a trust between two orgs is an enormous security commitment. If I had to do such a thing, I would spend a lot of time worrying about the security of the servers and desktops at the machine level, more so than the network level. To have the trust work at the network level requires opening a ton of ports, so to get good granular security, you need to work on the actual boxes.
I would strongly consider subnetting or isolating desktop machines. There really should be no reason that corp A's desktops need to talk to corp B's desktops. By blocking all such commo, you can reduce risk of spreading worms, virii,etc.
As part of the trust agreement, each org should get a test account in the remote domain. This should be a regular user account. You should spend a lot of time with this user account trying to see what you can access with it in your own organization.
Good luck
Matt
