If you have a critical application or if you are just going to be "picky" my advice is,
Don't use used capacitors. 
I would just measure the capacitance (there are multimeters that can measure capacitance)
and leakage resistance. In a power supply (or as an additional power supply filter), a meghom or so of resistance isn't going to matter so it's not that critical, but if it reads low the cap might be bad. And, there may be applications where 1M is too low.
I you haven't measured leakage resistance before... The meter will read low resistance when you first connect it and the capacitor begins to charge and the reading will then start rising and level-off at some point, or it may go beyond your meter's range and read infinity/open. (Non-electrolytics should read open.)
In general capacitors aren't critical. You can always use one with a higher voltage rating, and almost always use a higher capacitance value. A manufacturer can sometimes save money by using the same capacitor in multiple places on the board and simplifying the BOM, even though a variety of lower values would work. Often you could get-by with lower capacitance but that's not something I'd try "randomly".
Of course, the main exception is tuned filter circuits, including crossovers. Those should be within tolerance.
Electrolytic Capacitors do age, whereas most other capacitors, resistors, and other solid state electronics don't. I
think capacitors have a shorter shelf life than active life, but one datasheet I just looked at just says you should use a recovery procedure if it's on the shelf at 105 degrees C for more than 1000 hours... Not very long but I can't imagine ever hitting that kind of storage temperature. (It references a technical document that in a quick-search, doesn't appear to be available free.) The datasheet doesn't mention any limitations after that so I assume it's "good as new".
So...
Sometimes its large caps preservation to match the originals, stuff like that.
