That's a big question and I'm trying to decide how to answer it.
A lot depends, basically, on how literally you take the mathematics of Quantum Mechanics. The idea is that if a particle could, in principle, be in any state, then it makes sense to start by writing down the sum over possible states. If the possible states are 1, 2, 3 etc then the sum would be
current state = a*1 + b*2 + c*3 + ...
where the numbers a, b and c are weightings representing that it may be more likely for a particle to be in one state than another. The measurement will then pick out one of these states.
At this point is where the interpretations kick in. Is this sum physically real, or is it a mathematical trick?
If real, then the sum instantly implies that a particle is in multiple states simultaneously, until measured, when it will pick out one particular state with a certain probability (e.g. state 1 with probability a^2).
Since the states can also be positions, this also implies that a particle is free to be in several places at once until observed in one particular place, which is kind of freaky. Again, though, this depends on the "reality" of the current state. Since the numbers a, b, c can themselves not be "real" (often they include complex numbers, ie the square root of -1), then it's probably better to interpret the whole "current state" thing as "the square root of reality". Whatever the hell that means.
These are complicated things, anyway, and you can explain it in multiple ways. In the long run the most succinct answer is that "it just works", but you can have a lot of thing discussing the philosophy behind why it does.