Depression can be a profoundly disabling and costly disorder and is a major public health concern. Despite the efficacy of treatment options, it is often unrecognized, under-diagnosed, and inadequately treated in primary care settings. Research on patient-provider communication supports the connection among the quality of the patient-provider interaction, patient behavior, and health outcomes. The purpose of this study was to systematically examine the impact of patients' depressive symptoms on the patient-provider relationship, patient-provider communication, and patient satisfaction with the primary care office visit. One hundred twenty three patient-provider encounters were audiotaped and coded using the Roter Interaction Analysis System (RIAS). A 2 x 2 x 2 within-subjects factorial model provided the analytic framework for examining eight verbal communication behaviors categorized by speaker (patient or provider), type of utterance (question or information giving), and content of utterance (medical or psychosocial talk). Hierarchical linear modeling was used to analyze the two-level nested structure of the data. Results indicated that depression is associated with, but does not predict, increased provider-perceived difficulty in the patient-provider relationship. There was no significant change in either patient or provider communication behavior in relation to the severity of patients' depressive symptoms. Significantly more provider medical information was given during encounters with "difficult" patients and this behavior had a consistent negative effect on patient satisfaction. Patient-provider communication, by itself, does not appear to be a source of depressed patients' oft-reported dissatisfaction with medical care. Additional research is needed to further understand the core processes and structures of primary care practice in relation to the diagnosis and management of depression, their effect on patient outcomes, and to uncover opportunities for enhancing the effectiveness of depression care in primary care.