Objective: Communication skills are important, but the evaluation of these skills in the medical curriculum remains inconsistent. The aim of this study was to explore the relationship between the communication skills scores of junior medical students following curricula with different module evaluations and the clinical interview performance of the same students in a psychiatry department in their fifth year. Design: Prospective evaluation of communication and clinical interview skills. Setting: Medical school and psychiatry department inpatient setting. Method: The communication skills scores of 208 medical students in their second year (between 2009 and 2011) and their performance in interviews with patients in a psychiatry department in their fifth year (between 2012 and 2014) were assessed. Initial assessment was undertaken on the basis of problem-based learning (PBL) performance, evaluation of performance during conversation with patients, a related oral presentation and written report, students’ self-rating and a peer rating. Extra marks were given if students completed an optional special report inspired by the curriculum. To assess the interview performance of fifth-year medical students, the mini-Clinical Evaluation Exercise (mini-CEX) was used, together with weekly case-based discussions and the observation of a clinical interview with a real patient. Results: Tutors’ evaluations of the junior medical students’ performance during role-play ( ρ = .22, p = .001), its related reciprocal assessment ( ρ = .18, p = .010) and an extra-effort task ( ρ = .18, p = .009) as part of the communication skills curriculum were significantly correlated with the students’ later performance in a clinical interview with a patient. Conclusion: Observation of student interactions and motivations along with reciprocal peer evaluations offer better measures of communication skills in medical education than are oral presentations, self-ratings and conversations with real patients.