24 million people worldwide are affected by schizophrenia. Its complex psychopathology, including changes in perception, can incur substantial personal distress and economic burden. Finding appropriate treatment that attracts voluntary adherence is an ongoing challenge for clinicians to prevent relapse and poor prognosis. This thesis conceives the psychiatrist-patient alliance - mediated through talk - as an intervention point that demands analytic attention. Conceptualising ‘good’ communication is however hindered by a lack of a) conceptual clarity on its constituents b) knowledge of what actually happens in psychiatric encounters. Abstract ideals of ‘Patient Centredness’ and ‘Shared Decision Making’ are widely endorsed as beneficial to adherence, but do not pragmatically translate into specific practices, conducive to training. Following a preparatory systematic review, this thesis addresses a gap in literature by observing psychiatric communication in 3 mixed method studies. Synthesising coding methodologies and statistical analyses with principles of conversation analysis, two studies explore the association - and explanatory mechanism - between adherence and specific communication practices: patient other-initiated repair and psychiatrist questions. Treatment decisions, the precursor to adherent behaviour, are also examined: alternative resources that psychiatrists employ and their interactional consequences are mapped, with a focus on patients’ overt resistance. The findings collectively extend knowledge on medical interaction and demonstrate the utility of a novel approach to outcome research in field dominated by cross sectional studies. Clinical, methodological and theoretical contributions are yielded relating to six themes 1) the consequentiality of psychiatrists’ communicative choices 2) the manifestation of alliance and adherence in clinical talk 3) orientations to experiential expertise and the contingencies of antipsychotic medication adherence 4) reconceptualising ‘good’ communication: misalignment as key to clinical success 5) evidence of the interaction order in schizophrenia 6) reconciling the nuances of naturalistic interaction with global clinical outcomes.