You are right in saying that the voltage across the cell gives an indication on its charge status, but the problem is that that indication is relative to the specific model of the battery. For example, two AA cells both having, say, 1.35V don't need to have the same amount of energy stored. That depends on many variables related to the specific chemistry used by their manufacturer.
Therefore, although you won't damage two batteries with the same voltage reading, they will discharge at different rates, probably. After a while one of the two will have a much lower voltage and a correspondingly higher internal resistance and will act as an additional load for the other.
Probably the only safe approach is to use cells of the same model from the same manufacturer and built almost in the same period: manufacturers improve their chemistries from time to time and I won't trust two batteries to be identical if one has been built two years after the other.
As for the range of the difference in voltage, 1%-2% is probably safe, i.e. from a practical POV if their voltage differ only by a couple of digits in the third digits you are ok: 1.25 and 1.23 is OK, 1.25 and 1.29 is borderline, 1.25 and 1.35 is NOT ok.
Anyway, I recommend to do this mix only with low power gadgets, in order to reduce risks to abuse a cell. That is, if you draw very little current, that's ok. If you try to use that mix-up in a power-hungry device it could overstress the weakest cell in the set and cause leakage or something worse.
EDIT (prompted by a comment)
The source for what I said above is just personal experience. After all, you are asking something that is contrary to every industry best practice, thus I interpreted yours as an hobbyists' question and answered in that context. If you asked me for a professional advice about that I'd replied not to ever mix and match cells like that. No one in a professional context would do that.
Don't get me wrong, I also reuse old cells, both because of environmental concerns and to avoid wasting usable energy, but I rarely mix two old cells "having a different history". When I change batteries from an "high-power" gadget I measure the old ones, if they are still usable, say V>~1.1V, I put them away sticking them together with a rubber band, so that they don't mix with other old cells. Then, if I have a low-power gadget (wall clock, small LED flashlight, digital thermometer, etc.) that needs that number of cells I'm sure they have all the same level of charge.
In rare cases I have done the mix and match, measuring their voltage as I told you before. What's the rationale of my 1%-2% rule of thumb? If the difference in voltage is so low, I can safely assume (being the same model and coming from the same batch) their internal resistance is different roughly by a similar percentage. That means that the load unbalance when used in a gadget in percentage will be roughly of the same order of magnitude (a few percent). Assuming I use them in low-power gadgets, i.e. in devices that draw say 5%-10% of the cell's max current, the cell with higher resistance will dissipate a few percent more than the already low power average dissipation of each cell, which is well below the safe threshold, since I'm working with loads that draw much less than the max current a cell can stand.