McA's answer is technically correct, but going down the regexp route is likely to cause you more pain further down the road.

For example, have you considered HTML where a greater-than sign legitimately occurs in an attribute?

<td title="n > 5">n greater than 5</td>
[download]

Are you aware that the </td> closing tag is optional (as per the HTML 3.2, HTML 4 and HTML 5 specs). So the following is legitimate:

<tr>
  <td>1  <td>2  <td>3</td>
</tr>
[download]

You're better off using one of the many HTML parsing modules on CPAN which will already cover these corner cases.