Transient withdrawal of therapy has been advocated as a method of dealing with the complications of long-term use of levodopa in the treatment of Parkinson's disease. We retrospectively examined the effect of a 10-day period of levodopa withdrawal, or "drug holiday," in 28 patients. We then compared the subsequent clinical course of these patients over one year with that of 30 other randomly selected, similar patients with Parkinson's disease. In both groups the disease progressed; there was no difference in disease severity, capacity for daily living activities, or total amounts of dopamine agonists eventually used. For some patients, it was possible to reduce dopamine agonists used immediately after the drug holiday without causing deterioration, but a pulmonary embolus and other complications occurred. Subsequent complications related to long-term dopamine-agonist therapy during the follow-up period were similar in the two groups. this investigation indicates that a drug holiday carries some risk and does not improve the efficacy of levodopa therapy or prevent the problems that occur with long-term administration.