We propose a statistical analysis of immunity testing in EMC based on binomial distributions. This approach aims at extracting the immunity properties of a device from its probability of failure during a test. We show that under certain conditions, this approach can be applied to plane wave testing environments and reverberation chambers. This approach allows one to control the uncertainty of the immunity level estimation and to reduce the duration of a test by both reducing significantly the number of observations needed to reach a given uncertainty budget and giving an optimal number of power level tested. We show the benefits of such an approach for immunity testing and we present some experimental results.