XHTML is only useful if you want to autogenerate/manage/validate/etc the HTML code with help of a XML based tool, such as a component based MVC framework (e.g. Sun JSF, Apache Struts, Microsoft ASP.NET, etc) or with XSLT. Parsing/formatting HTML programmatically is trickier than XML, because HTML allows here and there non-closing tags, e.g. <br>. XML is much easier to parse/format programmatically because it is required to be well-formed.
If you're just starting and/or hand-writing "plain vanilla" HTML, I would recommend to use HTML 4.01 elements with a HTML5 doctype. There's really no need to massage the HTML code into a XML format.
<!DOCTYPE html>
<html lang="en">
<head>
<title>Page title</title>
</head>
<body>
<h1>Heading</h1>
<p>Paragraph</p>
</body>
</html>
The HTML 5 elements aren't widely supported yet, hence the recommendation to stick with HTML 4.01 elements. The HTML 5 doctype triggers the standards mode in most of the browsers, including IE6. The other benefit of HTML5 is that it allows closing shorttags like in XHTML. Also see HTML5 spec chapter 3.2.2:
Authors may optionally choose to use this same syntax for void elements in the HTML syntax as well. Some authors also choose to include whitespace before the slash, however this is not necessary. (Using whitespace in that fashion is a convention inherited from the compatibility guidelines in XHTML 1.0, Appendix C.)
Basically, even if you write pure XHTML, using <!DOCTYPE html> would still make it valid (and trigger webbrowsers in the correct standards mode).
