The address that you send from does not impact the abuse concern I raised. Let's say the purpose of this page is to e-mail copies of a jpeg to people who want it. As an end-user, I would go to your page, enter my e-mail address, hit submit, and moments later the jpeg is in my inbox. Sounds great. Now what happens when Joe Hacker writes a bot (trivial) which inserts 1_000_000 different people's addresses into that field? You've just allowed your server to be the source of spam.

I have no problem with asking questions. Please do. I'm clearly still answering them. But it's generally considered good form once you've established a dialogue on one issue to keep that within one thread. So, for example you might have asked MIME::Lite returns "SMTP Failed to connect to mail server: Bad file descriptor" in response to weismat's comment on Questions about sending e-mails. It means fewer people need to get up to speed, and it maintains a clear line of thought. In general, monks will check threads they've been involved with for at least 24 hours and generally longer, and many including myself have notification set up so I know when people respond to my nodes. It also keeps monks from repeating each other - for example, in writing this up, I found out that pileofrogs gave you the same security warnings in Re: Questions about sending e-mails. A little reading which may be of interest for you on this topic may be Re: php to perl and Re: Iterative Subroutine Approach Question. And if you think that your question is sufficiently off the original topic, you should include a reference to your previous thread like 5miller did in 752162.

Thanks a lot, kennethk. I agree with you. It is lack of time that tempts to ask a question in a new thread when I see that I need to rephrase it and can't wait to get the answer.
May I ask you one more question. Will it protect if I do the following.
First of all I do not allow attachements. All I need is to reply to the indicated email address with the link to my site with the generated ID. When it is clicked I got confirmation and register this user. When the email address is entered in the form I get it on the server side and make sure that there is only one there. Now, the bot may keep bombarding my form with addresses and pressing submit in the loop. But what I do is I do not accept submissions from the same IP with frequency higher than let's say 5 mins.
What do you think?

I do not accept submissions from the same IP with frequency higher than let's say 5 mins.

You cannot assume that a single IP address represents a single user. Perhaps with IPv6 that will be true, but it's not at all true for IPv4. A cookie or some other form of session data is much more reliable -- but beware that malicious clients may refuse to return your session data or may return stale but cached session data.

Those are all good things, and generally follow general good practice. The most important thing you didn't mention is that you not allow the client to modify the subject/body of the e-mail, since this would allow them to send out their advertisements. Part of this should include making sure that the address provided is actually just an e-mail address (see Code_injection). A good tool there would be Regexp::Common::Email::Address. If you are more paranoid (like me), you could also include a captcha, those funny looking character jpegs.