No theory worth having is going to have totally unambiguous applications to specific cases, and besides which there's no substitute for factual information and good judgment. That said, just saying we're going to take a prudent, empirical approach to questions turns out to not have any real content. In part, this is for formal reasons like "the interdependence of fact and theory" where people's empirical assessments of situations are influenced by their theoretical precommitments.
I often have difficulty explaining why it is I find noninterventionism to be uncompelling as a theoretical approach, but Matt explains it well. It's theory writ too large: every circumstance has the same answer, no matter the underlying conditions. Kosovo? Shouldn't invade. Rwanda? Shouldn't invade. The Sudan? Shouldn't invade. Taliban Afghanistan? Shouldn't invade. In fact, we should be content to let all those parts of the world do as they wish. But despite the fact that particular circumstances are effaced in the name of theory, history is brought in, in it's own haphazard way, to support the contention that interventions have never worked and, therefore, never will work. History, sovereignty, international law and national interest all come in, at various moments, when they support the predetermined answer, though they are often enough fungible between cases.
(I would be interested in a noninterventionist interpretation of the Mexican-American War, which I suspect would cautiously support the war and, moreover, definitely support the US territorial acquisition that resulted, but probably at the cost of logical inconsistency with, say, Kosovo).
By way of contrast, I don't mean to support forcible military intervention in all cases of human rights violations (though I am closer to the Altman and Wellman* position than, say, Daniel Larison's). Certainly one should aim at respecting the principle of non-intervention, which serves the important purpose of erecting some legal barriers preventing national-interest-sphere-of-influence logic, especially when former colonial territories are concerned. Though one respects the principle, it's important not to reify it: there are other operative parts of international law,** and serious arguments to be made about potential crystallization of customary international law around a limited right of intervention in some cases. Theory's job is to take the principles, moral, legal, and political, and the cases of intervention or failed intervention we have, and to attempt to derive conclusions to improve actions that are taken in the future. One thing this debate has made clear is that there's a wide consensus in America (and elsewhere) that intervention is justifiable at least some of the time, which for those of us who worried that the Iraq War might take interventionism off the table in American foreign policy, is welcome news indeed.
* "A Defense of International Criminal Law," Ethics 115 (October 2004): 35-67. Their basic position is that any violation of human rights can justify intervention. I take them to be on the far interventionist side of the spectrum. My own views would (not surprisingly) be closer to Allen Buchanan's idea of 'international legal reform,' though I think he relies too much on institutions to do the heavy lifting of political judgment.
**There are interpretive options available, such as the legal realist or New Haven School approaches to the UN Charter and collective security system, or the simpler appeal to progressive codification of human rights and the rise of the ICC/ICJ. One shouldn't be naive and adopt a progressive account of international law, but "international law" rarely speaks in one voice, let alone one which supports strong sovereignty and noninterventionism.