I don't see any mention of any encoding in this code, which is not good. And earlier you said: "I'm using the CGI module and have it properly set: print $q->header(-charset => 'utf-8');" so I doubt this code is representative.

You need to:

• Use a Perl version >= 5.12 and say use feature 'unicode_strings'; or use 5.012; (or higher).
• If you have any non-ASCII characters in your Perl script, save it as UTF-8 and add the use utf8; directive at the top.
• Make sure your data is coming from the database properly encoded. As I linked to above, you can check this via Devel::Peek. If you need that output to go to the browser, see this.
• Make sure you are doing binmode STDOUT, ':encoding(UTF-8)'; or use open qw/:std :utf8/;.
• Make sure you are telling your browser what encoding you are sending it.

Text::CSV is not the problem:

use warnings;
use strict;
use Devel::Peek;
use Text::CSV;

my $str = "\N{U+20AC}|\N{U+20AC}";
Dump($str);  # ... UTF8 "\x{20ac}|\x{20ac}" ...

my ($s1,$s2) = split /\|/, $str;
Dump($s1);   # ... UTF8 "\x{20ac}" ...
Dump($s2);   # ... UTF8 "\x{20ac}" ...

my $csv = Text::CSV->new ({ binary => 1, sep_char => "|" });
$csv->parse($str);
my ($c1,$c2) = $csv->fields;
Dump($c1);   # ... UTF8 "\x{20ac}" ...
Dump($c2);   # ... UTF8 "\x{20ac}" ...
[download]