We conducted a metanalysis of controled trials on the efficacy of levetiracetam (LVT) as add-on therapy for patients with refractory partial seizures. A MEDLINE search revealed five trials comparing the efficacy of different doses of LVT over placebo. Combined data confirmed the efficacy of LVT over placebo for the 50% reduction of seizures during therapy (x2=96.9; p<0,0001; OR=3.73; 95% CI=2.02-4.95), as well as for the complete resolution of seizures (x2=27.01; p<0,0001; OR=7.31; 95% CI=3.08-21.09). Only two trials showed a dose-related (1000 mg/d, 2000 mg/d o 4000 mg/d) efficacy of LVT. This suggest that the initial dose of LVT in these patients should be 1000 mg/d, and that the dose may be increased only in those who do not respond to therapy.