Hello there! 
First, are these balances used in an FDA setting, where they may require GLP compliance? If so, then your answer may take another approach where you need to demonstrate balance precision.
Having said that, the balance needs to be verified (for routine use) at the range it is used at. so if all you ever use the balance for is measuring product between 100-500 grams, then that would be your target range.
If you use your balance for one weight, and only one weight, then a single check is sufficient. However, rare is the case that a balance is only used for one measure. People from other labs, other areas could come in and weight product also.
Otherwise, you need to verify the balance across the range of use.
Now... tolerance is another matter. Like Howste correctly identified, Handbook 44 is a good document. However, most people with precision analytical balances tend to find that document.... not very helpful.
It may be best to establish your own tolerances based on process requirements. Also, be sure and document the performance checks and any noticed drift. That way you can establish how stable the balance is.
Finally, this should not take the place of having the balance calibrated by a qualified vendor. They can come in, remove error, perform corner loads, etc.

First, are these balances used in an FDA setting, where they may require GLP compliance? If so, then your answer may take another approach where you need to demonstrate balance precision.
Having said that, the balance needs to be verified (for routine use) at the range it is used at. so if all you ever use the balance for is measuring product between 100-500 grams, then that would be your target range.
If you use your balance for one weight, and only one weight, then a single check is sufficient. However, rare is the case that a balance is only used for one measure. People from other labs, other areas could come in and weight product also.
Otherwise, you need to verify the balance across the range of use.
Now... tolerance is another matter. Like Howste correctly identified, Handbook 44 is a good document. However, most people with precision analytical balances tend to find that document.... not very helpful.

It may be best to establish your own tolerances based on process requirements. Also, be sure and document the performance checks and any noticed drift. That way you can establish how stable the balance is.
Finally, this should not take the place of having the balance calibrated by a qualified vendor. They can come in, remove error, perform corner loads, etc.