Several models have been discussed in literature in order to evaluate the performance of the EDCA differentiation mechanisms. Despite of the exponential back off rules, usually the EDCA protocol has been modeled as a persistent slotted access protocol, and summarized into a set of access probabilities. These models often show a very different complexity and accuracy, according to the basis assumptions. In this paper, we propose a new modeling approach for EDCA, in which the medium access is described in terms of backoff counter comparisons, contention by contention, and no restrictive hypothesis is considered. Through our model, we verify the applicability of the most common protocol simplifications for different MAC parameter settings and load scenarios.