Clinical studies carry with them a paradox: The more obviously efficient an intervention is, the more ethical problems its trials pose. This article discusses the ethical problem of breached equipoise principle because of the perceived effectiveness of a nonblindable verbal technique, crisis dialogue (CD). CD is designed to help establish a therapeutic relationship with persons in a suspected psychotic state. In a pilot randomized controlled study in Yverdon, Switzerland (usual treatment vs. usual treatment + CD), after inclusion of 30 patients, clinicians expressed a consensual opinion that CD was effective in most cases. Following their opinion, the joint clinical and research team decided that the study had to be discontinued and that CD should be tried with all patients for ethical reasons. This poses an ethical problem with potential far-reaching consequences: In this interrupted study, differences between groups in terms of clinical outcome (Brief Psychiatric Rating Scale, Clinical Global Impression), therapeutic alliance (Working Alliance Inventory, Difficult Doctor–Patient Relationship Questionnaire), and patient satisfaction were consistent in favoring CD, but these differences did not reach statistical significance in most measurements. The early interruption of the study because of perceived effectiveness of the intervention can be seen as unethical as well because chances were high that a larger sample would have shown more conclusive results, allowing for faster introduction of CD in various clinical settings with corresponding improvement of patient care.