note Tanktalus <p>I'll second the anonymous monk's suggestion to use Text::CSV (or, better, Text::CSV_XS): <c>use Text::CSV_XS; use feature 'say'; use warnings; my @tests = ( [q[1|str|foo\\|bar|goo|2323] => [qw(1 str foo|bar goo 2323)]], [q[foo\\\\|bar] => [qw(foo\\ bar)]], ); for my $t (@tests) { my $s = $t->[0]; my @e = @{$t->[1]}; my $csv = Text::CSV_XS->new( { sep_char => '|', escape_char => '\\', } ) or die "" . Text::CSV_XS->error_diag(); $csv->parse($s); my @o = $csv->fields(); say "$s\n @e\n @o\n"; } </c> The regex/split suggestions fail to take into consideration a doubled escape prior to a delimiter (see the second test). For your example, that may be fine, but your example is so obviously contrived that it can't be taken definitively.</p> <p>Output: <c>$ perl5.16.2 x.pl 1|str|foo\|bar|goo|2323 1 str foo|bar goo 2323 1 str foo|bar goo 2323 foo\\|bar foo\ bar foo\ bar </c> Hope that helps.</p> 1002256 1002256