BACKGROUND: Subthalamic nucleus (STN) deep brain stimulation (DBS) of patients with PD allows reduction of antiparkinsonian medication but has only a mild direct effect on dyskinesia. Since antiparkinsonian medication has short- and long-term effects that may prevent an estimate of the maximal possible impact of STN DBS, such medication was used at the lowest possible dosage after DBS implantation. OBJECTIVE: To study the maximal and long-term effects of STN DBS using the lowest dose of medication. METHODS: Twenty consecutive patients with PD with motor fluctuations and dyskinesia underwent bilateral implantation under stereotactic guidance, microrecording, and clinical control. All medications were stopped before implantation and reintroduced, at the lowest dosage needed, only if the postoperative motor score did not reach the baseline level. Unified PD Rating Scale (UPDRS) motor (subscale III) scores were measured at baseline and after 3, 6, 12, and 24 months. RESULTS: After 21 plus minus 8 months, the UPDRS III "off-medication" score was decreased by 45% and was similar to the preoperative UPDRS III "on" score. Overall, medication was reduced by 79%, being completely withdrawn in 10 patients. Fluctuations and dyskinesia showed an overall reduction of >90%, disappearing completely in patients without medication. These improvements were maintained for 2 years. CONCLUSIONS: These results show that STN DBS could replace levodopa and allowed all antiparkinsonian medication to be discontinued in 50% of patients with PD. Fluctuations and dyskinesia disappeared completely in these patients but persisted in those still on medication. These improvements were maintained for 2 years.