You guys rock! Thanks!
No wuckers.
For example the Nichicon datasheet says no more than 4uA. Comparing tests.
In my understanding, the datasheet says that the leakage current is no more than 0.03CV or 4 (µA), whichever is greater for capacitors rated between 6.3 and 100V.
So, if you have, say a 470µF x 16V cap, the leakage current will be no more than 0.03 x 470 x 10⁻⁶ x 16 A = 225.6 µA. It will be equal or less.
But if you have instead a 4.7µF x 16V cap, using the formula will give you 2.246µA, which is less than 4µA. This means that in this case the leakage current for the rated voltage can be higher than 2.246µA but no more the 4µA.
Remember that you have to apply the rated voltage across the capacitor, not across the capacitor in series with the 10k resistor, and wait for at least one minute. The rated temperature also must be taken into consideration.
Will the same logic apply to ceramic and film caps?
Ceramic and film caps have other specs.
Kemet commercial grade high voltage ceramic caps have a leakage resistance of 1000/C (Ω) or 100 GΩ, whichever lower. So, for a 100nF, you should have 1000/(100 x 10⁻⁹) Ω = 10 GΩ (which is lower than 100 GΩ).
To measure it, you have to apply 500V across the capacitor for at least two minutes @ 25°C. If you use the 10k resistor technique you should read a voltage no more than 10kΩ x 500V/10GΩ = 500µV across the resistor (which corresponds to 50nA).