@layman,
It's important to understand that in this context, "accelerate" means something much wider than just "going faster". Any deviation in the direction of movement away from a straight line, any rotation, any change of speed, is a form of acceleration and makes a movement non-inertial.
Why all the fuss about inertial frames? Why do I say that the laws of motion as we know them only works in inertial frames? It's because they are the nearest thing you can get to a stationary frame. Let me explain in some length.
When one wants to predict the future movements of, say, pool balls on a billiard table, one needs a frame of reference to plot the initial position, speed and torque of the pool balls. In general, the billiard table itself provides a convenient frame of reference in this case. One will plot the balls in terms of their distances to the table sides, for instance. A pool player will do this unconsciously and intuitively, but a scientist will measure these distances as precisely as he can. As a result, they both can predict something like: If I can give the white ball this combination of speed, direction and torque, this and that will happen, and I can score big time...
Note that this assumes that the frame is inertial, ie in lay language, that the table is as stable and immobile as possible. Over the course of a pool shot, ie a few seconds, earth rotation does not matter, as the pool table moves almost in straight line over that short duration. Pool tables are usually located far from railway or subway lines, for instance. The subway tremor would "accelerate" the table up and down, introducing errors in the computation. Imagine that the billiard room -- instead of being in a quiet neighborhood, is bolted on the car on a large roller-coaster. The room now goes up and down and left and right and slows down and speeds up all the time... Thus the billiard table will not provide an inertial frame anymore. And sure enough, very odd things will start to happen to the balls, when viewed in that frame... Their behavior can no more be as easily predicted as before from within the room. They fly in all directions... A hazard really.
Does that mean that the laws of motion do not apply to the balls? No. But to describe the balls behavior in a way that fits the laws of motion needs an inertial frame, at a larger scale: a scientist modeling the entire roller coaster and the trajectory of the billiard table on it could try and predict the behavior of the balls. Good luck with that, but in theory it's possible.
When one realises that one uses a non-inertial frame, one needs to mentally 'step outside' that frame and find a better one, knowing there is no perfectly inertial frames.