Recently, it became fashionable to claim string theory is not scientific because it makes no testable predictions. Let me not again get into the testable vs falsifyable discussion, I still think people who still insist on falsifyabilty should quickly patch their philosophy of science module to a more recent than the 1934 version (Karl Popper's "Logic of Scientific Discovery"). But I hope everybody would agree that a theory can only be scientific if it makes statements about empirical observations.
Now, many people claim that string theory does not fulfil this criterion for a good theory. This view is fed by string theorists speculation about landscapes of vacua of which the number has not error bars in the exponent (like some numbers in astronomy) but in the exponent of the exponent. It seems that if string theory has so many different low energy versions that we can never extract any unique features that are observable.
In addition people become more and more excited about LHC taking data in a foreseeable future and there are many many different versions of signatures of supersymmetry or even low scale quantum gravity being visible in the new detectors. And they all look very different and once again it seems string theory can fit more or less every possible experimental outcome.
However, I think that many sceptics confuse two things here: The property that something can be observed in principle and that it can be observed in practice (now or soon or with today's technology). If we had a Planck energy accelerator, nobody would have doubts that strings could be observed. As soon as you probe distance scales that so smaller than the compactification scale, string theory look effectively 10 (or 11) dimensional and has quite unique properties.
Unfortunately, it is very very unlikely that LHC energies are at least of the order of the fundamental Planck scale so it is save to assume that LHC will not speak the last word about string theory. And it is as save to believe that we will never come even close to 10^19GeV with accelerators before the sun blows up. So it is likely that in practice, we will not be able to confirm (or rule out) string theory unless some surprise happens (like somebody finds a way to link cosmological observations rather directly to string theory).
But claiming some theory is not scientific because current (or foreseeable future) observational technology in practice does not make definite statements is not appropriate. This question should be decided based on whether observation is possible in principle.
Atomic theory is still scientific if you currently have only optical light frequencies at your disposal and you cannot resolve atoms directly. In that situation you are even luckier as quantum mechanics make definitive statements about absorption spectra for observable frequencies, still you cannot directly observe the atoms.
And I can still make scientific statements about the contents of the office across the corridor although it is locked and I don't have the key. In practice, I cannot test my speculations but in principle I could.