This website has compiled various regular expressions which try to resolve this problem. The editor's idea is great, using a set of valid/invalid emails, and a simple unit test, he can provide a good comparison of some of the most used regexps.

His philosophy is great : "It's better to accept a few invalid addresses than reject any valid ones, so I'm looking for 0 false-positives and as few false-negatives as possible."
But I've noticed 2 problems :

• His "best" regexp doesn't work in JavaScript (JS doesn't support advanced features like negative lookbehind ...)
• The method used to validate IP addresses is not correct (doesn't take care of 0-255 range)

So i've decided to improve an existant one, adding an other test criteria : also check the "real" validity of the IP address. The following work is based on the G. Arluison's improved version of Warren Gaebel's regex.

Here are my solutions :

/^[-a-z0-9~!$%^&*_=+}{\'?]+(\.[-a-z0-9~!$%^&*_=+}{\'?]+)*@([a-z0-9]([-a-z0-9_]?[a-z0-9])*(\.[-a-z0-9_]+)*\.(aero|arpa|biz|com|coop|edu|gov|info|int|mil|museum|name|net|org|pro|travel|mobi|[a-z]{2})|([1]?\d{1,2}|2[0-4]{1}\d{1}|25[0-5]{1})(\.([1]?\d{1,2}|2[0-4]{1}\d{1}|25[0-5]{1})){3})(:[0-9]{1,5})?$/i

This one works very well (found 18/18 valid mails + deep IP address check, and found 19/20 invalid mails - there is a problem checking global length)

There's just a small problem, each time a new TLD > 2 chars will be added, you'll need to append it to the list in the regex, if you want a more generic solution, you can use this variant (note that this version will not check if the TLD really exists) :

/^[-a-z0-9~!$%^&*_=+}{\'?]+(\.[-a-z0-9~!$%^&*_=+}{\'?]+)*@([a-z0-9]([-a-z0-9_]?[a-z0-9])*(\.[-a-z0-9_]+)*\.([a-z]{2,6})|([1]?\d{1,2}|2[0-4]{1}\d{1}|25[0-5]{1})(\.([1]?\d{1,2}|2[0-4]{1}\d{1}|25[0-5]{1})){3})(:[0-9]{1,5})?$/i

Those 2 solutions should be usable in all languages providing PCRE (Perl Compatible Regular Expressions), on server & client side (such as Javascript, PHP, Perl, Python, Ruby etc...)