We tested both Rogers's hypothesis that listening enables speakers to experience psychological safety and our hypothesis that the benefit of listening for psychological safety is attenuated by avoidance‐attachment style. We tested these hypotheses in six laboratory experiments, a field correlational study, and a scenario experiment. We meta‐analyzed the results of the laboratory experiments and found that listening increased psychological safety on average but that the variance between the experiments was also significant. The between experiment variance in the effect of listening manipulation on psychological safety exposes a methodological challenge in choosing a research paradigm of good‐versus‐normal listening, as opposed to normal‐versus‐poor listening. More importantly, we found, as expected and across all designs, that the higher the avoidance‐attachment style was, the lower the effect of listening on psychological safety. This finding has implications both for practice and for placing a theoretical boundary on Rogers's theory.