Background: Generating alternative interpretations is a widely used technique in cognitively-oriented treatments for psychosis, including social cognitive treatments. However, research on judgment and decision-making suggests that this technique may backfire among individuals with cognitive deficits. Aims: Mary/Eddie/Bill (MEB) was developed as an alternative approach to help improve Theory of Mind, attributional bias and Jumping to Conclusions in psychosis. An initial pilot trial was conducted to evaluate the tolerability and potential efficacy of MEB. Method: Twenty-four outpatients with psychotic disorders participated in an uncontrolled trial. Social cognition, social functioning, and participant feedback were assessed. Results: Treatment completers exhibited statistically significant improvements in Theory of Mind, social cognitive overconfidence and self-reported social engagement, and reported positive experiences with MEB. No improvements were observed in attributional bias. Conclusions: Given that there are risks associated with the use of traditional generating-alternatives techniques, the current findings support ongoing development of MEB and similar approaches.