In a realistic scenario, not very far, maybe just a few miles if you're discussing fragments large enough to leave impact craters. A fair bit further than that if you're talking smaller debris. The distance depending on how high the meteor broke apart and how fast and at what angle it was approaching.
If you consider that the dense part of the atmosphere is thinner compared to the earth than the skin of an apple, see here, then it's logically unlikely that the atmosphere could scatter any impacting object across the entire earth.
Theoretically, you could probably work out an ideal scenario where the meteor is vulnerable to split very high in the atmosphere at a nearly flat approach where half the meteor skips on the atmosphere a few times and the other half doesn't and they end up hundreds of miles apart, but such a scenario is unlikely based on the usual "break apart" altitude being measurably lower than the "skip or bounce across the atmosphere" altitude.
This speed in atmospheric flight puts great pressure on the body of a
meteoroid. Larger meteoroids, particularly the stone variety, tend to
break up between 7 and 17 miles (11 to 27 km) above the surface due to
the forces induced by atmospheric drag, and perhaps also due to
thermal stress. A meteoroid which disintegrates tends to immediately
lose the balance of its cosmic velocity because of the lessened
momentum of the remaining fragments. The fragments then fall on
ballistic paths, arcing steeply toward the earth. The fragments will
strike the earth in a roughly elliptical pattern (called a
distribution, or dispersion ellipse) a few miles long, with the major
axis of the ellipse being oriented in the same direction as the
original track of the meteoroid. The larger fragments, because of
their greater momentum, tend to impact further down the ellipse than
the smaller ones.
As an Earth-grazer passes through the atmosphere its mass and velocity
are changed, so that its orbit, as it re-enters space, will be
different from its orbit as it encountered Earth's
There is no exact criterion for passing by outside of the atmosphere,
except perhaps roughly 80 km (50 mi) up, or the Kármán line at 100 km
The Chelyabinsk meteor that hit Russia in 2013 approached at high speed and a low angle, breaking apart higher than usual.
On account of its high velocity and shallow angle of atmospheric
entry, the object exploded in an air burst over Chelyabinsk Oblast, at
a height of around 29.7 km
and that left a debris field roughly 17 miles (based on the map and the longitude/latitude coordinates). Source. The larger bits, each a few KG or so were mostly within a 10 mile range. For a larger meteor splitting in two pieces large enough to leave impact craters, as your question focused on, the altitude would likely be lower as the larger body should hold together longer and retain more of it's velocity and direction towards the Earth. There would be less dispersion.
The Tunguska event, for example, is thought to have broken up much closer to the earth.
the meteor is thought to have burst in mid-air at an altitude of 5 to
As a final bit of info on this, the double meteor impact in Sweden is thought to be two meteors orbiting each other that impacted earth together. The impact craters happened at the same time and are roughly 16 KM apart. It's unlikely that one meteor of that size could have split in two large parts, each leaving impact craters that far apart, so the more likely scenario in that case is that it was two meteors orbiting each other not one meteor that broke into two pieces.
A final point on this, a single meteor breaking up into two pieces that both leave impact craters is probably rare. Meteors will often break up, but usually into many pieces, not two. Chelyabinsk, in the link posted above, did break up on two parts initially, in it's large blast, but those two pieces each broke up further before reaching the ground. I'm not saying it can't happen that one meteor will leave two impact craters, but it's probably a statistical rarity.
As a 2nd sidebar, if it breaks apart flying inside the Roche Limit, like Comet Shoemaker-Levy did flying around Jupiter before impact, then the impacting pieces can spread across much of the earth, but that's not possible with an atmospheric break-up.
(Corrections welcome, as I'm a hobbyist, not an prof. astronomer)