The first question is how often the weights are used. The next question is how they are stored and how they are handled. These two things will generally drive that answer.
A weight used weekly or daily, I would say needs to go on an interval not more than one year, and less is preferred.
If gloves or tweezers are not used to handle them, then the weight changes the minute they are picked up, because of finger oils.
Weights used once or twice a year, stored in a locked cabinet in a controlled environment, and always handled wearing gloves or using tweezers, I would say can be pushed up to five years.
A weight used weekly or daily, I would say needs to go on an interval not more than one year, and less is preferred.
If gloves or tweezers are not used to handle them, then the weight changes the minute they are picked up, because of finger oils.
Weights used once or twice a year, stored in a locked cabinet in a controlled environment, and always handled wearing gloves or using tweezers, I would say can be pushed up to five years.
The key to deriving the optimal re-calibration interval is the data. However, until you can generate enough history to make decisions, I would go with Hershal's recommended intervals.
It's important the users understand how much they can affect mass standards by improper handling.