The objectives were to evaluate gabapentin add-on therapy in a large population under conditions close to real practice and to determine the therapeutic doses as reached with adaptable dosages. A 6-month multicentre, open-label study, involved addition of gabapentin to pre-existing treatment at the initial dosage of 1200 mg and subsequent adjustment between 900 and 2400 mg/day according to efficacy and tolerability. A study group of 610 adult patients, with partial epilepsy, persistent seizures and a median seizure frequency with a baseline of 7.2 per month were recruited; one-third had less than four seizures per month. Polypharmacy was frequent, with a mean of 2.3 concomitant drugs. After 6 months, 368 patients (62%) continued on gabapentin, at a mean dosage of 1739 mg/day with 44% of responders. On an intention-to-treat basis, median reduction in frequency was 21.2%, and the responder rate was 33.9%. The responder rate increased to 40.7% in the less severe subgroup receiving only one concomitant drug. Seventy-nine patients (13.4%) remained without seizures during the last evaluation period, versus nine (1.5%) during the baseline. Most of them had initially less than four seizures per month. The most frequent adverse effects, somnolence (29.3%), asthenia (14.6%), nausea (7.9%), ataxia (7.7%) and vertigo (7.2%), occurred rapidly after initial titration to 1200 mg/day, and were usually transitory. Weight gain (8.8%) seemed to be related to gabapentin dose. The combination of two recent drugs, vigabatrin and gabapentin, in 190 patients led to similar efficacy levels, with a tendency for more frequent somnolence and asthenia.