The reason for your confusion is because of the simplistic, but erroneus way this topic is often taught in a school class. I was not even satisfied with what my standard med school physiology texts taught me about membrane equilibria, so I had done some research then. Of course, I've lost touch, but was able to refresh myself quickly with a bit of googling.
Read thru' these two links VERY carefully.
http://human.physiol.arizona.edu/SCHED/CV/Wright/14membra.htm
http://human.physiol.arizona.edu/SCHED/CV/Wright/15action.htm
The short version : the "real" reason for the resting memb pot of any cell is NOT the sodium-potassium pump (at least not directly). The real reason is something known as the Nernst equilibrium. Essentially, if you have a concentration gradient of a single ion between two sides of a semipermeable membrane that exhibits high permeability to that ion, you will get diffusion of that ion down its chemical gradient. But at the same time, that act of diffusion is accompanied by a gradually accumulating electrical gradient that opposes the movement. A point is reached where the electrical gradient will exactly counterbalance the chemical one, this point is called the Nernst equilibrium (NE). At the NE, no net ionic flux occurs and the membrane potential is stable. The NE is dependent on two things, the permeability of the membrane to the ionic species and the initiating chemical gradient. Interestingly, with a high chemical gradient and high permeability, the ionic flux required to swing a membrane from neutral to the NE potential is so slight that the original internal and external concentrations can be taken as not having altered. A lot of understanding of this whole thing hinges on that point : the actual concentrations are not really changing much at all. Rather, it is the *tendency* of movements to occur (which depends on changing permeabilities) that governs membrane potential.
The Nernst equation allows us to calculate the NE point for different ionic species. The combined equation for Na, K and Cl is the Goldman-Hodgkin-Katz equation. These two are explained in the first link I provided.
Essentially, for a resting cell, the memb pot is close to the Nernst potential for potassium. This is because the resting cell is far more permeable to K than it is to Na or Cl. The sole function of the active Na-K-ATPase is to establish the chemical gradient for K (and incidentally Na) with high K inside and low K outside. The natural leak of the K from inside to outside following this (due to high permeability to K) is responsible for causing the resting membrane potential. Just a small leak (relatively) is sufficient to establish the potential, as I explained earlier. If the pump is shut off and all else remains constant, the cell membrane will not go immediately towards neutrality because this is a stable electrochemical state (OTOH, the cell will take on water and lyse of course).
So it's now easy to explain what's going on with nerve cells. There are K channels always working in the cell that maintain the memb pot near the Nernst pot for K. A stimulus opens Na channels that moves the cell closer toward the Nernst pot for Na (which is positive). At a threshold pot, more voltage gated Na channels open up and the process is reinforced, the cell depolarises. Slower opening voltage gated K channels now start opening while the Na channels start closing, and the cell starts going back toward the K Nernst potential. At this point, the memb is even more permeable to K than in the resting state (because both the nongated and gated K channels are open) so the cell goes even further toward the K Nernst potential, resulting in the afterhypolarisation. The repolarisation toward resting memb pot is due to closing of the voltage gated K channels so that only the nongated K channels are open. The cell is now at resting memb pot.
Note the distinct lack of involvement of the active pump in the process of the AP and the aftermath - that's only involved in any residual cleanup and maintenance of osmotic pressure.
