We report the results of a double-blind, randomized prospective trial on D2 and 5-HT2 receptor occupancy and the clinical effects of olanzapine versus clozapine in a sample of neuroleptic-refractory schizophrenic patients. Receptor occupancy was evaluated in different cortical areas and in basal ganglia using [18F] fluoro-ethyl-spiperone ([18F] FESP) and positron emission tomography (PET). A total of 15 neuroleptic-free patients completed the study undergoing a baseline and a post-treatment PET scan (olanzapine, nine patients, one female; clozapine, six patients, three female) 8 weeks after starting treatment. PET data were analysed both by regions of interest and on a voxel-by-voxel basis using Statistical Parametric Mapping (SPM96). Olanzapine and clozapine induced a similar and significant inhibition of [18F] FESP binding index in the cortex. In the basal ganglia, receptor occupancy was significantly higher with olanzapine than with clozapine (p=0.0018). By contrast, no differences in receptor occupancy were detected at the level of the pituitary gland. Clinical outcomes, in particular a full extra pyramidal tolerability, were similar. In this sample of neuroleptic-refractory schizophrenic patients, olanzapine and clozapine showed a different pattern of occupancy of D2-like receptor despite a common lack of extrapyramidal side-effects.