Objective: The authors quantitatively examined differences in psychiatric residents' and attending physicians' communication profiles and voice tones. Methods: Audiotaped recordings of 49 resident-patient and 35 attending-patient medication-management appointments at four ambulatory sites were analyzed with the Roter Interaction Analysis System (RIAS). Nonparametric tests were used to compare differences in proportions of speech devoted to relationship-building, activating, and partnering in decision-making processes, and data-gathering/counseling/patient education. Differences in affect expressed by psychiatrists' voice tones were also examined. Results: Residents' visits were twice as long as Attendings' visits (28.2 versus 14.1 minutes), and residents devoted a significantly greater proportion of their talk to relationship-building (23% versus 20%) and activating/partnering (36% versus 28%) aspects of communication, whereas Attendings devoted a greater proportion to biomedically-related data-gathering/counseling/patient education (31% versus 20%). Analysis of voice tones revealed that residents were perceived as sounding significantly friendlier and more sympathetic, versus Attendings, who were rated as sounding more dominant and rushed. Conclusion: These findings show distinct communication profiles and voice-tone differences. Future psychiatric communication research should address the influence of appointment length, psychiatrist/patient characteristics, and other potential confounders on psychiatrist-patient communication. (Contains 1 figure and 2 tables.)