Abstract. Classical models in strong field physics typically require a careful selection of initial conditions which correspond accurately to the state selection. We compare two such procedures, the (many) pilot atom technique and the microcanonical strategy. We show that for some range of parameters, the pilot atom strategy may lead to artifacts due to the correlations between the different initial conditions. On the other hand, the microcanonical strategy is, by design, free of such artifacts for all admissible parameters.