Another, equivalent way to interpret the results is that drawing 100 samples from the truely-random chi-square distribution should result in around five values that are at least as large as Vo.
If the chi-square distributions drawn from empirical tests produce values Vo a ratio higher than 1 out of 20, is seems safe to conclude that the number streams being used by the empirical tests are not random. And the higher the ratio over 0.05, the safer you are in your conclusion.
This page last modified on 2 March 2005.