If the output is really 4/5 HTML with a smattering of perl, the eperl program works nicely. This automatically generates the Content-type headers. Anything that's inside a "special" HTML tag <? !> is taken to be perl, while everything outside is taken to be HTML. The shebang line changes to something like #!/usr/local/bin/eperl -mc. Eperl calls the perl compiler/interpreter on the machine to handle the perl parts. Supposedly you can get it to work with mod_perl as well, though I haven't tried that myself.