Not only that, but one should clarify either XHTML transitional vs. strict. The OP asked what is the "safest" to use, but did not clarify specifically what qualifies or quantifies "safe".
To me, that means the least amount of problems rendering the code cross browser compatible. To others it might mean the easiest to learn. To still more it might mean the least security vulnerabilities. To yet more it might mean simply the one where over time the code will remain compliant.
There is no single answer to all these, but I will say this much (my .02):
XHTML 1.0 strict technically is an XML application and has the toughest syntax, it involves COMPLETE separation of style from content, and one needs to learn not only XML and CSS in depth, but also deal with a few variances (i.e. "id" is used in anchor tags, not name attribute just to name one). All tags must be properly nested, closed, in lower case and have only one root element. If one uses the correct DTD at the top of their code, follows all the rules and validates the page properly the user will benefit from less HTML and increased cross browser rendering support. Meaning, the site will look the same across most browsers if not identical, darned close since XHTML 1.0 strict standards are well defined and well supported.
The transitional form is a mixture with loose syntax between both the HTML 4.01 and XHTML worlds and is easiest to learn and the easiest code to port from HTML.
HTML 4.01 is just fine for most these days for most uses, honestly. I'd suggest learning XHTML transitional and then progressing to strict for jobs that require adoption of the toughest standards for ALL of the reasons I stated above when I described "safe".
All are related so the learning curve is not incredible, one just needs to scan over the
XHTML W3C specification and maybe reference syntax through the
W3C schools site. (actual links provided)
Remember, there is alot more to this than what I mentioned - this is just a general overview and my .02, by no means the law of the land.
-jim