ssh hanging, but *not* ServerAliveInterval

Question

This is a new problem, but I have no idea what has brought it on.

I have two local clients (Windows/Cygwin and Linux/SL6.6, on the same IP address behind NAT), both making ssh connections to the same remote server (Centos 5.6), for the same user.

ssh always works perfectly from the Cygwin client, but it has started hanging on the Linux client. It's slow to connect to the remote server and, if it does connect, it hangs after a few seconds. If I type in 3 or 4 commands in quick succession (uname -a, for example), then the local terminal (urxvt or konsole) hangs half-way through the command output. I can get back out of ssh with the normal ENTER ~ . sequence.

Other info:

1. On the SL6.6 client, I have a .ssh/config file which sets ServerAliveInterval to 240 for all hosts
2. Both the local clients (Cygwin and SL6.6) use key authentication; the SL6.6 client still has this problem with password authentication
3. yum update openssh says that the SL6.6 client and the server are both up-to-date
4. The server never auto-updates, but the SL6.6 client might potentially have auto-updated something that I didn't notice
5. The SL6.6 client produces debug error messages when trying to authenticate with gssapi, but ssh -o GSSAPIAuthentication=no doesn't help
6. /var/log/secure shows exactly the same mesage whether I log in from Windows or Linux: same user, same IP address (different port), 'accepted publickey'
7. Load average 0.00, 0.00, 0.00

since the same user is logging in from Windows or Linux, the standard answers (see here, for example) don't seem to apply. Any ideas? Thanks.