Jeff Hoffman suggested the following code, that's quite a bit cleaner than what I came up with, and it also works quite well. <dtml-call "REQUEST.set('z_t', _.string.split(_.str(ZopeTime()))[0])"> <dtml-call "REQUEST.set('u_all', 0)"> <dtml-call "REQUEST.set('u_logged', 0)"> <dtml-call "REQUEST.set('u_today', 0)"> <dtml-call "REQUEST.set('u_not', 0)"> <dtml-var u_all> <dtml-var u_logged> <dtml-var u_today><br> <dtml-in count_users_logged> <dtml-call "REQUEST.set('u_all', u_all + 1)"> <dtml-if logged_on> <dtml-call "REQUEST.set('u_logged', u_logged + 1)"> <dtml-if "_['z_t']==_.string.split(_.str(_.DateTime(logged_on)))[0]"> <dtml-call "REQUEST.set('u_today', u_today + 1)"> </dtml-if> <dtml-else> <dtml-call "REQUEST.set('u_not', u_not + 1)"> </dtml-if> </dtml-in> I'm a bit mystified that the only detectable difference between this version and *many* that I tried without success looks to be the spaces between objects and keywords in expressions, along with the complete absence of <dtml-let>. e.g <dtml-call "REQUEST.set('u_logged', u_logged + 1)"> vs. <dtml-call "REQUEST.set('u_logged',u_logged+1)"> I've looked at lot's of code samples that have no spaces, and have seen enough error messages to think very hard about random spaces in expressions. This is a good example of how code lore (superstition) gets proliferated, but it's hard to argue with working examples. Thanks again, to Jeff, Jim, Pavlos, et. al. Jerry S.