We report the results of a collaborative study on malignant syndrome (MS) that developed in patients being treated with levodopa and other anti-parkinsonian drugs. We analyzed clinical features, laboratory findings, precipitating events, and risk factors for poor outcome. The study was conducted in five centers in Japan. Patients who developed MS between January 1991 and December 1997 were included. The enrollment criteria used were the same as those for neuroleptic MS proposed by Levenson et al. (1985).A total of 99 episodes were encountered in 93 patients (72 with Parkinson's disease and 21 with secondary parkinsonism); one patient had four recurrences of MS and three patients had two recurrences. High fever was the most frequent clinical manifestation of MS followed by worsening of parkinsonism, and then altered levels of consciousness. Serum creatine kinase was abnormally elevated in all the patients studied. Life-threatening complications were rhabdomyolysis, disseminated intravascular coagulation, and acute renal failure. The most frequent precipitating event was discontinuation or dose reduction of anti-parkinsonian drugs, particularly levodopa. No drug was the exception in the precipitation of MS. Intercurrent infection was the next most common precipitating event. MS developed without drug withdrawal or infection in some patients. In five patients, severe "wearing off" phenomenon was the only event preceding the onset of MS. Hot weather and dehydration appeared to be the cause in three patients. Among the total of 99 episodes, patients recovered to the pre-MS state following 68 episodes (68.7%); in the remaining 31.3%, patients failed to recover to their previous state. Older age, higher Hoehn and Yahr stage during the symptomatic phase of MS, higher akinesia score, and the absence of wearing off phenomenon prior to developing MS were associated with poor outcome. The most frequently used treatments of MS were intravenous fluid, levodopa, dantrolene sodium, and intragastric bromocriptine. Early introduction of treatment is important. Any elevation of body temperature during the course of anti-parkinsonian drug treatment should be considered as MS until proved otherwise.