Three hundred and forty seven patients with epilepsy from 54 centres across Europe not fully controlled with sodium valproate (VPA, n = 117), carbamazepine (CBZ, n = 129), phenytoin (PHT, n = 92) or phenobarbital (PB, n = 9) monotherapy were recruited into a lamotrigine (LTG) substitution study. If 50% or more seizure reduction occurred (responders) on addition of LTG, an attempt was made to withdraw the original antiepileptic drug (AED). If successful, this was followed by a 12 week period of LTG monotherapy. Overall, 73% patients completed the add-on phase (47% responders), 41% attempted AED withdrawal and 23% achieved LTG monotherapy. In the 60 patients (17%) completing the trial by remaining on LTG monotherapy, median monthly seizure frequency was reduced from 6 during baseline to 1.7. Sixteen percent of patients were withdrawn due to adverse effects, mostly during the add-on phase. Dizziness and diplopia occurred most frequently in the CBZ group, nervousness and ataxia in the PHT group, and rash and tremor in the VPA group. Slower LTG dose escalation resulted in fewer withdrawals due to rash in the VPA-treated patients (38% to 8%, P < 0.01). The responder rate was higher (P < 0.01) in patients with idiopathic tonic-clonic seizures (61%) than in those with partial seizures (43%). The addition of LTG to VPA (64% responders) produced a significantly better response (P < 0.001) than adding it to CBZ (41% responders) or PHT (38% responders). This effect was seen for partial (VPA, 57%; CBZ, 39%; PHT, 39%; P < 0.02) as well as tonic-clonic seizures (VPA, 70%; CBZ, 53%; PHT, 50%; NS). These data lend credence to the suggestion of therapeutic synergy between LTG and VPA.