tbook's (X)HTML files will almost always be validated by web validators. However, sometimes the W3 validator claims that attributes in tbook's HTML4 output contain invalid characters. This is a limitation of XSLT 1.0 processors, and it only occurs if you use non-ASCII characters in ID and URL attributes. However, it is not possible to use XSLT 2.0 processors at the moment, for various reasons.
tbook prints a warning message in such cases. May may choose a less dangerous attribute value. On the other hand, those validation errors are absolutely nothing to worry about anyway.