OBJECTIVES: To examine psychiatrists’ preferences for specific attributes of antipsychotics in the treatment of schizophrenia.   METHODS: A discrete choice experiment (DCE) was administered to psychiatrists to examine 7 attributes of antipsychotics (efficacy, mode of administration, formulary access, onset of action, dosing frequency, safety, and side effects). Psychiatrists were presented with 18 total choices and asked to indicate which of 2 hypothetical medications they preferred. Choices were divided among 3 profiles of patients with schizophrenia (average patient, symptomatic nonadherent patient, and chronic nonadherent patient). Descriptive statistics were reported for psychiatrists’ sociodemographic and professional characteristics. Part-worth utilities were estimated using random effects logit models, and relative importance values were calculated for the attributes.   RESULTS: Complete data were available from 478 psychiatrists; their mean age was 52.1 (±9.4 SD) years, and the majority were male (n=326; 68.2%) and Caucasian (n=354; 74.1%). The psychiatrists had a mean of 19.0 (±9.1 SD) years’ experience practicing psychiatry. Across all patient profiles, the efficacy attribute consistently had the highest relative importance (RI): 54.93%. Mode of administration (RI=13.51%) and formulary access (RI=11.33%) also contributed notably to the psychiatrists’ medication preferences. Other attributes were of more minor importance, each with RI values <10%, including onset of action (RI=6.95%), dosing frequency (RI: oral=6.23%; injection=0.94%), safety (RI=4.30%), and side effects (RI=1.80%). The RI of medication attributes showed some differences across patient profiles; mode of administration increased in importance for both types of nonadherent patients, while formulary access and safety decreased in importance.   CONCLUSIONS: The results of the DCE suggest that efficacy is the most important factor for psychiatrists’ making medication decisions regarding the treatment of patients with schizophrenia. The RI of efficacy does not vary by patient profile; however, the RI of other attributes tends to vary depending on the profile of the patient being treated.