Clozapine, an atypical neuroleptic agent, improves dopaminergic-induced psychosis in parkinsonian patients without increasing motor disability. However, because of the risk of agranulocytosis periodic hematological controls are mandatory. Olanzapine, another atypical neuroleptic, does not require such monitoring, which may represent a practical advantage. Therefore, for 12weeks we compared the tolerability and efficacy of clozapine and olanzapine in two groups of nine consecutive parkinsonian psychotic patients treated with these compounds. All the patients on clozapine (mean starting dose: 13.1+/-7.9mg/d) completed the study despite reporting a number of adverse events, including somnolence, falls, orthostatic hypotension, and syncope. In contrast, early withdrawal occurred in three of the nine patients receiving olanzapine, due to severe gait deterioration and drowsiness (mean starting dose: 3.9+/-1.3mg/d). Psychotic symptoms improved in both groups, as reflected by a reduction of 71.7% in clozapine and a 61.7% reduction in olanzapine, in five selected items from the Neuropsychiatric Inventory. On conclusion of the study, parkinsonism had improved in the clozapine group with a 19.7% decrease in the raw scores and a 7.9% decrease in the weighted scores according to the Cornell University Rating Scale for parkinsonism (mean dose: 16.9+/- 10.3mg/d). Conversely, the six patients receiving olanzapine who finished the study experienced aggravated parkinsonian symptoms, with a 25.5% worsening in the raw scores and a 24.6% worsening in the weighted scores (mean dose: 4.7+/-2.3mg/d). We postulate that the early drop-outs in the olanzapine-treated parkinsonian group may be attributable to a non-specific effect of the drug as a result of starting at too high a dose, and that the worsening of parkinsonism following prolonged treatment may have been caused by the drug's blocking effect on striatal D2 receptors.