The introduction of levodopa in the 1960s revolutionised the treatment of Parkinson's disease (PD), and it continues to be the most effective symptomatic therapy. The vast majority of PD patients who start treatment with L-dopa experience good to excellent functional benefit. In vitro studies with high doses of L-dopa and absent glia had shown that it may be neurotoxic, but other tissue culture studies show L-dopa to be neuroprotective. Most studies in animal models and clinico-pathological and mortality studies in humans failed to show evidence in favour of accelerated dopaminergic neuronal loss with long-term L-dopa therapy.L-dopa continues to be beneficial throughout the course of PD, although as the disease progresses, escape of some symptoms from adequate control may occur and refractory disabilities such as impaired balance, dysarthria, cognitive decline and hallucinations may emerge. Treatment of advanced PD may also be complicated by the emergence of motor fluctuations and dyskinesias. Studies in animal models and in humans show that these motor complications are not specific to a particular dopaminergic agent, but that they are related both to the extent of the striatal lesion and to the mode of application of dopaminergic agents: Pulsatile administration of L-dopa and of the dopamine agonist apomorphine causes more motor complications than continuous striatal dopaminergic receptor stimulation, and continuous administration can alleviate existing dyskinesias and fluctuations. Several controlled studies comparing levodopa and dopamine agonists as initial treatment have attempted to answer the question whether delaying L-dopa therapy can reduce the occurrence of motor complications. Three medium-term (3-5 years) and one 10-year study showed less dyskinesia in the first five years of treatment in patients who had started therapy with a dopamine agonist. However, these studies also consistently showed that levodopa provided better functional improvement in the first years of treatment. Ten-year follow-up data in patients randomised to L-dopa or bromocriptine also showed a slightly lower incidence of motor complications in the bromocriptine arm. However, this difference was not significant for the clinically relevant moderate and severe forms of dyskinesias and fluctuations, and was achieved at the expense of significantly worse disability scores during the first years of therapy. Furthermore, the relative impact of motor disability and dyskinesias on patients' quality of life remains to be established. Concordance is essential in the optimum treatment of PD and patients should be informed of the various therapeutic options available. Treatment should respect individual patients' needs and take into account their particular functional disabilities and specific handicaps. Low-dose L-dopa therapy (up to 400 mg/day), however, remains the most effective initial treatment of choice for the majority of patients.