The study of cerebral variables associated with response to neuroleptics holds interest from both theoretical and clinical points of view. To date, no studies have aimed to identify predictors of response to olanzapine based on cerebral measurements. Here, we used magnetic resonance to assess the relationship between volumes of the prefrontal (dorsolateral and orbitofrontal) and temporal (temporal lobe and hippocampus) cortical regions and ventricles and, on the other hand, the response to olanzapine in 16 schizophrenic patients. Data from 42 healthy controls were used to calculate volume residuals in the patients, defined as deviations from the expected values, given individual age and intracranial volume. Residuals thus represent the effect of illness on regional measurements. The association between clinical change and those residuals was calculated separately for the positive, negative and total scores from the Positive and Negative Syndrome Scale (PANSS). There was a significant direct association between the degree of orbitofrontal atrophy and the improvement of positive symptoms with olanzapine. No predictors were found for change in the negative dimension. A trend was found for patients with larger ventricles to show a greater global decrease in total PANSS scores. Neither age nor duration of illness explained a significant proportion of the symptom improvement. This result, together with others from the literature, supports the idea that atypical antipsychotics may offer some benefit to patients with significant regional atrophy, and this may have implications for the choice of antipsychotic in clinical practice.