Empirical research has repeatedly proven the facilitating effect of “natural frequencies” when it comes to solving Bayesian problems (e.g., Gigerenzer & Hoffrage, 1995). However, in a study with 183 university students, we found that the beneficial strategy of translating complicated conditional probabilities (e.g., “80%”) into natural frequencies (e.g., “8 out of 10”) is for the most part not being used by participants. On the contrary, even when statistical information was presented in natural frequencies, most of the students translated the natural frequencies they had been given back into probabilities – with the unfortunate consequence that they could no longer solve the task.