The only information one has about a
particle (or a system for that matter) in quantum physics is its
wavefunction.
The wavefunction is related directly to the probability distribution of the particle. In other words, the wavefunction allows one to find out the probability that the particle will be found in a given place if you were to measure its position. The shape of the wavefunction itself is harder to figure out; the
Schrodinger Equation can be nasty to solve.
When one actually performs the
measurement (in this case, position), one gets a specific location as a result. If you redo the measurement immediately afterwards, you should measure the same position again.
This means that the wavefunction of that particle has
collapsed. In other words, instead of a complicated
function with lots of peaks and valleys, the wavefunction is now a sharp spike at the location where you measured the particle. The
probability of finding the particle there is now 100%, and all information about what the wavefunction was is now gone.
Let the
system sit, though, after the first measurement, and due to the
uncertainty principle, the wavefunction begins to creep outward again, slowly... until you again have a
probability distribution, with no sure knowledge of where you'll find the particle. Back to square 1!
Collapsed wavefunctions don't stay that way