Here's one way. Note though that for any significant amount of HTML parsing you really really should use one of the HTML modules such as HTML::TreeBuilder or HTML::Parser.

use strict;
use warnings;

my $text= "<td class=sp></td><td>Hello</td><td class=sp></td><td>Bye</
+td><td class=sp>";
my @matches = $text=~ m|<td>((?:(?!</td>).)*)</td>|g;

print "@matches";
[download]

Prints:

Hello Bye
[download]