According to Bell’s theorem, no local realistic model can reproduce all predictions of quantum mechanics for the EPR-B experiment. Any such model would have to predict both the perfect correlation and breaking the CHSH inequality, but this seems impossible, since to provide the first prediction, the model would have to be deterministic, whereas this would seem to make the second one impossible. A model of the EPR-B experiment is presented, in which this apparent contradiction is expected to be avoided due to a deterministic chaotic mechanism underlying measurements. The model is in the phase of a “black box” model, since this mechanism is not yet specified, but only a corresponding probabilistic response function.