Harrie Verveer's Blog: Benchmarking Xdebug
Normally this tool would be used on your development environment, or in rare occasions on some remote environment that is similar to the production environment. [...] Since you are adding extra overhead it is very likely things will run slower. But how significant is this difference. Would it be something worth considering?
To lay the ground work, he describes the hardware he's running the tests on and the software he used - PHP 5.2.6 (Suhosin) on Debian with Xdebug 2.1.0 - and the script he used to run the tests. Several tests were run multiple times (3) showing quite a difference in performance when using the debugger versus not. He also suggests a more practical testing scenario with WordPress and ApacheBench. The numbers still support the other results.
In a side note, the Twitter account for Xdebug has mentioned some updates that have been made to the version currently on their SVN server that can help with some of these performance issues.