A single dose of levodopa (L-DOPA) reduces motor disability in Parkinson's disease (PD) for a few hours, a short-duration effect. However, there are suggestions that L-DOPA may also produce a long-duration benefit of some days. In the present study, we examined the long-duration action of L-DOPA by observing the time taken to achieve maximum stable benefit after starting a constant dose of sinemet-CR (sinemet-CR) (200 g L-DOPA/50 mg carbidopa) twice daily in nine newly diagnosed patients, and the time taken to deteriorate back to baseline after stopping treatment. A single dose of sinemet-CR (200 mg L-DOPA/50 mg carbidopa) had little obvious short-duration action on the Unified PD Rating Scale (UPDRS) motor scores in the majority of patients, either before starting chronic sinemet-CR therapy (200 mg L-DOPA/50 mg carbidopa, b.i.d.) or after chronic treatment. However, all patients gradually improved on chronic sinemet-CR therapy, taking 9.3 +/- 1.8 days to achieve maximum response. On stopping chronic sinemet-CR treatment, it took 6.8 +/- 3.0 days for the same patients to deteriorate back to baseline motor disability. In similar experiments, the time taken to deteriorate back to baseline after stopping treatment with the directly acting dopamine agonist ropinirole (9-21 mg daily) in eight other de novo patients with PD was found to be 6.2 +/- 1.7 days. The long-duration effect of L-DOPA and ropinirole may, therefore, be due to some slowly evolving postsynaptic pharmacodynamic change in the central nervous system (CNS). Loss of this long-duration action may be responsible for the emergence of motor fluctuations on chronic L-DOPA therapy.