SSH 1 through 3, and possibly other versions, allows local users to bypass restricted shells such as rbash or rksh by uploading a script to a world-writeable directory, then executing that script to gain normal shell access.
SSH Secure Shell for Servers and SSH Secure Shell for Workstations 2.0.13 through 3.2.1, when running without a PTY, does not call setsid to remove the child process from the process group of the parent process, which allows attackers to gain certain privileges.
Buffer overflow in the URL catcher feature for SSH Secure Shell for Workstations client 3.1 to 3.2.0 allows remote attackers to execute arbitrary code via a long URL.
The default configuration of SSH allows X forwarding, which could allow a remote attacker to control a client's X sessions via a malicious xauth program.
ssh 2.0.12, and possibly other versions, allows valid user names to attempt to enter the correct password multiple times, but only prompts an invalid user name for a password once, which allows remote attackers to determine user account names on the server.
SSH server (sshd2) before 2.0.12 does not properly record login attempts if the connection is closed before the maximum number of tries, allowing a remote attacker to guess the password without showing up in the audit logs.