This question is rarely, if ever, asked but assumptions regarding the length of time involved regularly pop up, especially when measurement disputes are the situation into which such assumptions are injected.
Everyone operating a calibration laboratory encounters this phenomenon when a report they have issued is being questioned. Many customers assume that the issued report will apply until the gage or instrument involved is next up for calibration. This is not an unreasonable assumption until you consider the outrageous calibration frequencies some customers apply and, sadly, many quality auditors will accept.
 I have seen customers that have their working gage block sets with a ten year cycle or production gages on a five year cycle. Despite what they imply, these are not always guesses out of thin air. Usually they have nothing to do with metrology but are derived from budgetary considerations. Quality requirements usually come in second compared to cost cutting.
ISO 17025 is quite clear about such matters. The owner of the equipment being calibrated is the one who calls the shots when it comes to calibration frequency. Who else knows the conditions of use, or anticipated future usage? Certainly not calibration labs who may tend to do the opposite to the customer: recommend a shorter period to drum up business. Some customers insist that their calibration reports must show a due date possibly because there is an air of credibility to it when it is on a document issued by an outside party. As you can see from these comments, the laboratory is simply listing what the customer tells them to and is not in a position to endorse due dates at all. Occasionally, the customer’s suggested due date is so far off the wall the calibrating laboratory notes it has been provided by the customer to preserve the lab’s reputation.  
In a perfect world, a calibration report will be valid throughout the calibration cycle, but as you can see, the time span of some cycles stretch the limits of metrology, physics, metallurgy, tribology and who knows how many other disciplines. When the cycles are all over the map, the time validity of a calibration report will be also.
In a practical world, the only time a report can be considered valid is while the calibrated items are in the lab. Once they leave this “safe haven” all bets are off.
The reason for this is the different environments they end up being used in, and the extent of the use and abuse they are subjected to. Even generalized rules don’t apply very well unless a number of elements are equal and they rarely are.  
It could be argued that a fixed limit gage won’t change all that much after checking a thousand parts based on calibration results and thus a report will be good for at least a thousand parts. This logic assumes nothing changes over the next thousand parts. But if it is used on more abrasive materials or is used by an abrasive operator or the work is not as clean as it was, this logic goes down in flames.
Like the financial statements produced by accountants, calibration reports are a snapshot in time of the state of a gage or instrument. Neither is suitable for predicting the future but a review of an item’s history as told by its calibration reports over time can be helpful in this regard.
If the person who sets the calibration cycles is kept in the loop regarding future production plans, a cycle may need to be shortened by the knowledge that a second production shift is ahead. Similarly, when the calibration report shows a gage is close to a rejection limit, the cycle may be shortened so it remains in limit while in use.
All of this requires someone to study the calibration reports. I know this is pretty radical thinking but sooner or later someone has to do it otherwise useful information will become useless.
To recap: A calibration report is valid for as long as the calibrated item remains in the laboratory. Once it leaves, the working world will take its toll and usually invalidates the report.