I've used HttpUnit, and it's reasonably good.  Now I'm trying HtmlUnit.

One advantage seems to be Xpath support.  Looking at [[http://htmlunit.sourceforge.net/apidocs/com/gargoylesoftware/htmlunit/html/xpath/HtmlUnitXPath.html|the javadocs]], it appears that you can get the document node(s) indicated by an Xpath expression as
{{{
 XPath path = new HtmlUnitXPath("a/b/c");
 List results = path.selectNodes(domNode);
}}}
From there it should be easy to make assertions about the matching node(s).