You are using a relative path. The current directory is not what you think it is. Your script dies before printing valid HTTP headers. Always print out valid headers first. Or use CGI::Carp with the fatalsToBrowser option.