Internal standards are great because they allow you to compare those two chart printouts. If you didn't have an internal standard, how do you know the temp hasn't changed? How do you know some nearby power-line wasn't turned on and pumped in a bunch of noise to the machine? A multitude of things could have gone wrong between the time you ran your initial sample and your final sample.
In a real research lab, a GC-MS will probably have many people using it if your samples are run hours apart--so who knows what they did to it? Maybe they peed in the mobile phase (sorry that's LC).
If you run internal standards each time, and THOSE peaks come out to be the same, then you know your good to go and you can compare the peaks from your two readouts, if not...who knows if something happened that effected the 2nd reading.