Our experience with atypical antipsychotics in patients with PD is that their motor effects are not predictable. The multiple reports concerning clozapine's beneficial effects on tremor, dystonia, nocturnal akathisia, and dyskinesias all underscore this observation. However, the appearance of even minor degrees of parkinsonism in normal volunteers or schizophrenics should suggest that an antipsychotic will not be well-tolerated in patients with PD. The treatment of PD is probably the most stringent test of a drug's freedom from parkinsonian side effects. The data from trials in schizophrenia concerning parkinsonian effects cannot always be confidently interpreted. Virtually all subjects in these trials have been treated with typical neuroleptics until shortly before study entry. Because the parkinsonian side effects of these drugs may persist for several months, patients may still show declining levels of parkinsonism even when placed on a drug that induces it if this effect is milder than that induced by the pre-study neuroleptic. Depending on the pre-study drug used and the duration of the study, distinguishing placebo from a low-potency neuroleptic may be impossible. Furthermore, the standard measure of parkinsonism in psychiatric studies is the Simpson-Angus scale which is heavily weighted toward rigidity and may underscore bradykinesia, gait, and posture abnormalities. The prolactin response to an antipsychotic drug may turn out to be a good predictor of its freedom from parkinsonian side effects. That would fit with the data presented above of clozapine and quetiapine having less parkinsonian effects, olanzapine having more but variable effects and risperidone being poorly tolerated. With the data presented above, comprising a current review of all reports of the use of atypical antipsychotics in PD that we could locate, we can say little with certainty. The only drug with confirmed benefit without worsening parkinsonism is clozapine. Open-label trials involving over 400 patients and two multicenter, placebo-controlled, double-blind trials have demonstrated that it is effective in treating the psychosis. It improves tremor, does not worsen other motor functions to any significant extent, and is safe at low doses. Limited data provide conflicting information on both risperidone and olanzapine. Quetiapine seems to be well-tolerated with some, but definitely less, worsening of PD motor features than risperidone and olanzapine. Based on the current literature, our personal experience, and anecdotal experience of other PD specialists which we solicited, we will venture our own interpretation and recommendations. We think risperidone is poorly tolerated and should be used only as a last resort; that olanzapine is better than risperidone but will, in a majority of patients with PD, worsen motor function. We are optimistic, but not yet convinced, that quetiapine may prove to be as effective and better tolerated than clozapine. It will not require cumbersome monitoring because it does not induce a blood dyscrasia. We therefore recommend that DP be treated in the following manner. First, the anti-PD medications should be simplified and reduced as much as tolerated. We think, in general, side effects multiply more with increasing numbers of drugs than with drug dose, so that patients are more likely to tolerate a higher dose of levodopa than a lower dose of levodopa combined with other adjunctive anti-PD medications. In reducing anti-PD medications, we recommend tapering and stopping, if necessary, the drugs with the highest risk-to-benefit ratio first. Anticholinergics are stopped first, then selegiline, dopamine agonists, amantadine, and finally COMT inhibitors, which have no psychotomimetic action of their own. Finally, levodopa is reduced. Generally, a point is reached at which the anti-PD medications cannot be reduced without jeopardizing motor function. If psychosis persists at this point, then an antipsychotic is added. (ABS