A number of client and server implementations of SSH version 1 have been found to contain a potentially serious security problem. There is some evidence that this problem may be exploitable in practice in at least some implementations. If you are running an implementation which is subject to this problem, you should contact your vendor for an upgrade as soon as possible.
The following documents have further information about this flaw:
Note that this flaw is a different and potentially much more serious problem from the "session key recovery" flaw which was publicized around the same time.
Here are the SSH ports I have been made aware of. I have not evaluated most of the packages listed below, and their appearance on this list should not be interpreted as a recommendation or endorsement. Commercial products are listed separately at the end; to my knowledge, all of the other packages listed here can be downloaded and used by individuals for non-commercial purposes free of charge. Consult the documentation accompanying each software distribution for further information regarding their licencing.
Update: When SSH Communications Security introduced ssh-1.2.29 to correct security problems in the software (and no other features added), they modified the license agreement in such a manner that many users who previously qualified to use the product without paying license fees cannot legally do so with the new version. I consider this practice to be of highly questionable ethics.
They've played such games with their license agreements in the past, and this has been progressively eroding the goodwill they had built up with the community over the years. If you're one of the people who SSH Communications Security is apparently trying to extort into purchasing licenses in order to address a security problem, I would strongly recommend that you consider switching to a competing product with a friendlier license and greater respect for their user community, such as OpenSSH.
February 2001: SSH Communications Security has begun engaging in the dissemination of fear, uncertainty and doubt regarding mostly theoretical vulnerabilities in version 1 of the SSH protocol--or, should I say, the SECSH protocol, since they also assert that referring to the protocol as "SSH," or including those letters in the name of a software package which implements it, amounts to an infringement of their trademark. One can hardly blame them for wanting to protect their business model, but the erosion of goodwill continues.
Please report broken/outdated links to djast@cs.toronto.edu; thanks.