Information on the long term effects of lamotrigine (LTG) in children is scarce. We report on the efficacy and tolerability of add-on LTG for 144 weeks in 13 children aged 4 to 14 years with refractory epilepsy. In the first 4 weeks, LTG at a dose of 2.5 +/- 1.9 mg/kg (mean +/- SD), and trough serum steady-state concentrations of 2.6 +/- 1.5 mg/L, reduced seizures by > 50% in only 3 patients. When LTG was increased in week 12 to 5.7 +/- 3.4 mg/kg (6.1 +/- 3.9 mg/L), seizures were reduced by > 50% in 6 patients (completely in 2). From week 12 to week 144, 5 patients withdrew from the study because of inefficacy or seizure increase, and one withdrew at the parents' request. In the remaining 7 patients, LTG at a dose of 9.1 +/- 7.1 mg/kg (5.3 +/- 2.6 mg/L) reduced seizure frequency from a median of 5.0 seizures/month at baseline to 1.4 seizures/month in week 144 (p < 0.03 according to Wilcoxon's test), a seizure reduction > 50% in 4 patients (with complete control in 1 patient). Four of the 13 children showed adverse events (3 behavioural and 2 tremor), but none developed skin rash. The LTG concentration/dose ratio was significantly lower in the 25 samples from children taking phenytoin or carbamazepine (0.32 +/- 0.11) than in the 21 from children receiving valproate (3.5 +/- 1.3), and was intermediate in these taking phenytoin or carbamazepine + valproate (1.4 + 0.2). In conclusion, long term treatment with add-on LTG at serum concentrations of 5 mg/L may be useful in about 30% of children with refractory epilepsy, but a 10-fold interpatient variability in LTG concentration/dose ratio should be taken into account when LTG is associated with other antiepileptic drugs.