In three experiments, Daryl Bem’s experimental method on retroactive facilitation of recall was replicated, with minor modifications, with the aim of testing robustness of Bem’s retroactive effect. In Experiment 1, the original doubleblind procedure was replaced by the single blind procedure under which the experimenter, but not participants, knew what the experimental conditions were. In the experimental condition of this experiment Bem’s results were successfully replicated, however, in the control condition, which was supposed to produce null effect, the effect was significant and a mirror image of the effect in the experimental condition. In Experiment 2, the control condition was run before the experimental condition. As in Experiment 1, the results in both conditions were symmetrical, though this time not significantly different from zero. In Experiment 3, in the experimental condition the results were reversed to those in Experiment 1. Altogether, the results marginally support robustness of Bem’s reported effect. However, the results do not support the interpretation of this effect as an effect in inverted causality. Rather, the results of this study suggest that the effect reported by Bem is a particular case of a number of possible outcomes, which occur when Bem’s method is applied. The results also suggest that these outcomes occur due to a direct effect of the observer’s mind on the RNG functioning.