Some here hinted at the fact that they might not have a guarantee that they keep that accuracy over time - but those two factors don't necessarily have anything to do with each other.
Also keep in mind that unlike what some people seem to think, the 1% is not a guarantee that every resistor will be within 1%, but rather an indication of what the standard deviation is (I think 2 sigma but I'm not sure). This means that most will perform (significantly) better than 1%, but some will perform worse - it's just very unlikely.
This also brings me to another point about resistor tolerance: Sometimes we see people put 10 1% resistor in parallel "to make a .1% resistor" (it actually also works in series). This only works if you can assume that all the tolerances are uncorrelated. If this is not the case, and two resistors that are next to each other in the reel are not uncorrelated, then you will not get the desired tolerance - in fact it may be worse.