In the abstract, of course it's a bad idea, for all the other valid reasons (e.g. "storing it in plaintext", "interception", "account hijacking and replay", etc.) that other respondents have stated.
The issue that I'd like to raise, however -- as is nearly always the case with IT security-related questions like this -- is, "compared to 'what'?" If an alternate method, less vulnerable to interception (preferably, out-of-band, e.g. ".zip your password into an encrypted archive, send it to you via e-mail and then call you by voice and dictate the password to you") method is available, then obviously that should be used.
But what if no such alternate channel is available, or is practical?
I can make a case that if the password that an end-user is likely to pick is "password" or "123456", that is inherently far more vulnerable than having a complex password e-mailed to that user, with instructions to use it only for a limited time and then change it to something equally complex (or to store it in a password manager like KeePass, LastPass or Password Safe, and then delete the plaintext version)?
Unless your name is "Snowden" or "Assange", the relative chance that all your communications will be constantly monitored -- and therefore you are at risk of having the TCP/IP data-stream intercepted and thereby having your password intercepted -- is actually very low. Whereas, the risk that a weak password will eventually be hacked, is quite high.
So it all comes down to a situation of risk analysis. If you are in a high-threat environment (for example you are a bank, and you need to communicate account credentials to high-income customers), then, yes, ANY form of password communications that can be intercepted, is likely not appropriate. If, on the other hand, you are running the Peoria, Ohio, Ladies' Auxilliary Floral Arrangement Society on-line discussion forum, and you need to give members password-protected access to the "Post Your Latest Flower Power Combinations" forum, then the consequences of a successful password interception and subsequent on-line attack, are likely to be much less, and perhaps a greater level of risk in communicating the credentials, can be accepted.