Let me try and answer this from an almost newly minted CSSLP's (pending endorsement) perspective.
The principle of complete mediation would say no. However, the principle of psychological acceptability might say yes. It really depends on what the token allows you to do and the risks those actions carry.
Does it allow:
Users to launch nuclear missiles
Users to access their credit card information they have saved in an account profile
Users to post messages on a cat pictures blog
You'd probably not want to reuse any tokens if they give you authorisation to launch nuclear missiles you'd want this to require authentication each and every time.
You might decide you can relax the authentication requirements a little for access to credit card information so it authenticates the first time but then can reuse the token a few times before requiring re-authentication and set a shorter timeout than the default 30 mins.
While you might even choose to extend the 30 mins timeout if some harsh comments about your cat pictures is the worst you will face.
Also, it depends what protections you have in place concerning other security aspects of your system. If the rest of the security is not up to scratch (commensurate with the system's criticality) then token reuse might be the least of your worries.
If using for identification such as OAUTh or SAML I don't see why reusing it would be an issue. In fact this is what it is about, SSO. The key is that the process support integrity of the identification. If so, it should be used frequently and often
Life is short... Eat more Cookies! Brian Hildebrandt CPA, CISSP