Context The language and structure of doctor-patient consultations have been widely researched. Although simulated patients (SPs) are much used in communication skills teaching, the language of the simulated consultation has not received much attention. Objectives This study aimed to resolve the following questions. How are interruptions and numbers of words distributed in simulated consultations? Do they correlate with set variables (e.g. gender, scenario) or outcome variables (e.g. grade)? Methods A total of 100 videotaped assessed consultations between SPs and Year 3 medical students were transcribed. Words by each participant were counted, and interruptions were coded and counted. Amount of talk and interruptions were chosen because they are potential markers of conversational dominance. Results We found that SPs talk (55% versus 45% for students) and interrupt (74% versus 26% for students) significantly more than medical students. The scenario is significantly associated with the number of words and interruptions. Multivariable testing shows that female SPs are associated with more words. The number of words is significantly and positively associated with examination grade. The number of student interruptions is significantly and positively associated with grade. Conclusions Because the simulated consultation takes place in an institutional setting, the SP may have institutional power over the student. This may explain how findings from these role-play interactions differ from actual doctor-patient consultations. This suggests that simulated consultations are educational devices rather than exact representations of doctor-patient interactions. The authors hope this paper will contribute to a discussion about the nature of role-play in medical education.