People who cannot control their epilepsy with medication should be referred sooner for surgery, suggests a study in The Lancet.
The University College London study found that 47% of patients who had surgery were free of seizures after 10 years.
It tracked 615 adults annually over an average of eight years.
But experts say people should be realistic about the risks and benefits of this kind of surgery.
Surgical treatment for epilepsy is being used more and more.
Two-thirds of people with epilepsy have the condition well controlled with medication and experience no side-effects.
The other third, for whom medication does not work, could be eligible for surgery.
The most common type of surgery undertaken on people in the study was temporal lobe surgery, which focuses on the area of the brain behind the forehead between the ear and the eye, where most seizures originate.
Researchers who carried out the study, published in The Lancet, reported that 63% of all patients were free of seizures two years after surgery (excluding simple partial seizures), 52% after five years and 47% after 10 years.
Those with simple partial seizures (SPS) in the first two years after temporal lobe surgery were two-and-a-half times more likely to experience subsequent seizures than those who experienced no SPS.
This has implications for decisions to stop taking anti-epileptic medication, say the authors.
They found that a relapse was less likely the longer a person was free of seizures after surgery, and more likely the longer seizures continued after surgery.
Realistic about risks
Most patients who were seizure-free after surgery still chose to remain on an anti-epileptic drug of some kind.
John Duncan, medical director at the Epilepsy Society and professor of neurology at University College London's Institute of Neurology, which carried out the study, said the study gave a realistic view of the long-term outcomes of surgery for epilepsy.
"It means people should consider surgery sooner rather than later, while being realistic about the risks too."
The study says that the selection process for surgery and surgical methods need to improve to increase success rates and to more accurately identify those who will not benefit from surgery.
Prof Duncan said the threshold for surgery had come down over the past 20 years.
"Surgery used to be offered to someone having three or four seizures a month. Now it could be offered to someone having one seizure a month, who is in employment but who is experiencing side-effects from the drugs.
"So the stakes are higher. The balance of risk and benefit has changed as surgery is offered to those less severely affected."
Commenting on the findings of the study in The Lancet, Dr Ahmed-Ramadan Sadek and Professor William Peter Gray from the Wessex Neurological Centre at the University of Southampton said the study validated the long-term effectiveness of epilepsy surgery.
But there were still hurdles to be overcome, they said.
"The median duration of epilepsy before surgery in this study was 20 years. In view of the long-term results of surgery shown, clinical practice needs to change with the early referral of appropriate patients."