That's a big question! Seriously, I assume I'm talking to an audience of scientists here, and that we all more or less understand the nature of approximations and experimental error. I assume we understand that all truth has error bars and that our knowledge is imperfect.
I consider Newton's Laws to be "true" even though I'm fully cognizant of the modifications that must be made to them to account for relativity and quantum mechanics. I suspect (though I don't know for sure) that some modifications to QM are yet required before gravity will get incorporated. Nonetheless, I don't think this makes QM untrue, even potentially.
John Flanagan asked me to elaborate on fuzzy logic. Since I've got that on my mind, it occurs to me that a fuzzy description of truth might be more appropriate here. There are degrees of truth, and those degrees can move relative to the circumstances. (Newton's laws are slightly more true for the orbit of Jupiter than they are for the orbit of Mercury, for example.) Nonetheless, an approximate, contingent, or imperfect truth is not the same as a falsehood.
One of the big mistakes made by many external critics of science is that they look at isolated incidents and mistakes, and jump to the conclusion that because science has and will sometimes come to the wrong conclusions, that therefore scientific truth is no such thing. They're making the same mistake they're accusing the scientists of. They're assuming that the correctness of science is a yes or no question, that if science is wrong it can't be right; that there are no gradations to truth. Scientists understand the limits to what they say. Experimental papers that don't include error bars are routinely rejected. I'm not sure the liberal arts understand the importance of error bars.
Even in the social sciences, you often find people quoting standard deviations and other statistical measures simply because they know it's expected, but with no real understanding of what it means or why it's important. It's just a number that comes out of another intrinsic function in the spreadsheet. (Note: there are many social scientists who DO understand statistics. Not coincidentally, these don't tend to be the ones who spend their time criticizing science. People who understand and use statistics, in whatever field, are precisely those people who have adopted the scientific epistemology and world view.)
I guess I am something of a neo-Platonist. I do believe there is an objective reality. I do believe that this reality adheres to certain laws. I do believe in the principle of induction, and that we can use experiment to achieve better understandings of and approximations to the way the universe really is. And I also believe that as long as we recognize that our "laws" are approximations subject to experimental revision in new areas where they've yet to be tested, that they are in a very real sense true.
Finally, one footnote. To learn more about these issues and why they're important you should read the various material written about the Sokal hoax. You may also want to read Carl Sagan's The Demon Haunted World.