Decreased gray matter volume (GMV) in the superior temporal gyrus (STG) has been implicated in the neurophysiology of schizophrenia. However, it remains unclear whether volumetric reduction in the subregions of the STG can predict treatment efficacy for schizophrenia. Our cohort included 44 drug-naive, first-episode patients, 42 unaffected siblings and 44 healthy controls. Voxel-based morphometry and pattern classification were utilized to analyze the acquired imaging data as per the anatomical subdivision by a well-defined brainnetome atlas. The patients presented lower GMV values in left TE1.0/1.2 (TE, anterior temporal visual association area) than the siblings, and lower GMV values in the left/right TE1.0/1.2 and left A22r (rostral area 22) than the controls. A positive correlation is observed between the GMV values in the right A38l (lateral area 38) and baseline Positive and Negative Syndrome Scale (PANSS) total scores in the patients. Support vector regression (SVR) results exhibited a significant association between predicted (based on the GMV values in the right A38l) and actual symptomatic improvement based on the reduction ratio of the PANSS total scores (r = 0.498, p = 0.001). Our results suggest that normal structure in the right A38l of the STG may be an important factor indicative of the effects of antipsychotic drugs, which can be potentially used to monitor drug effects for first-episode patients at an early stage in clinical practice.