Crosstalk noise evaluation consists in analysing the effect of the transition of a signal - called aggressor - on its neighbours - called victims. This evaluation is based on a model that estimates the peak noise produced on each victim. However, from a victim's point of view, not all its aggressors are able to make a transition at the same time. Thus, determining the active aggressors of a victim at a given time is a key point to get a realistic estimation of the noise. We present a static analysis method for crosstalk noise evaluation. The purpose is to calculate the "maximal noise configuration" that is likely to occur for each signal. The proposed approach is based on an event-driven algorithm. First, the instability periods of signals are calculated. Then, using these periods, the maximal noise configuration is obtained for each victim.