By breakdown, you may be at it a while. I've heard of 6.3V caps shorting at 120V+. Probably varies a lot from part to part, too.
Capacitance change with bias is a poor indicator. By that measure, a 1uF 0805 might be only 16V while a 0.1uF 0805 (same material) might be solid to 50V or more. Same material, same construction, just more and thinner layers. Which is good to keep in mind, in general, really.
If you have the means of measuring dissipation factor, add that to the list. Higher K materials (with codes like Y5V and Z5U) are very poor (>10%), medium K (X7R, Y5P) are reasonable (1-10%), and low K (C0G) are very low loss (< 0.1%).
On the upside, C0G are almost always white, porcelain in appearance, while high-K types are yellow, orange, tan or brown sort of colored.
Tim