Getting older teenagers to lead sex lessons for youngsters has no effect on their having unprotected under-age sex, researchers say.
About half the study group had conventional, teacher-led lessons
In an experiment, students aged 16 and 17 gave lessons to 13 and 14 year olds in English schools.
It was thought safe sex messages would carry more weight coming from people the younger children could relate to.
Fewer girls went on to have sex before they were 16 - but there was no effect on the boys' behaviour.
The UK has the highest under-18 pregnancy rate in western Europe, with 90,000 pregnancies a year.
Researchers at University College London and the Institute of Education carried out the study among 8,000 pupils from 27 schools.
About half had the "peer-led" sex education from older students who had been given training and support, while the others received conventional, teacher-led sessions.
The researchers said similar topics were addressed. The biggest difference was in skill-based activities such as practising putting a condom on a penis-shaped object in the peer-led groups.
Subsequently, "significantly fewer" girls from the peer-led lessons reported having intercourse before the age of 16 - 35% compared with 41%. There was no difference among boys.
And the rates of unprotected first intercourse - the main interest of the study - were the same, at about 8% of girls and 6% of boys.
Fewer girls became pregnant, 2% against 3%, but the researchers said the numbers involved were too small to draw firm conclusions at this stage.
The peer-led classes were popular, although more than half of girls and about a third of the boys said they would have preferred single-sex lessons.
The lead researcher, UCL's Judith Stephenson, said: "Getting older teenagers to teach the younger ones about sexual health and relationships could be a step in the right direction.
"Further follow-up to age 20 will provide a fuller evaluation of the effects of sex education on sexual behaviour and pregnancy.
"We now have a good idea about the elements that pupils themselves think make a good school sex education programme.
"However, we should also continue to look at other ways of educating and supporting young people outside of school."
One of the findings was that nearly half of young people learnt most about sex outside school.
The results were published in the medical journal The Lancet.
In an accompanying commentary, an Australian expert - Roger Short from the University of Melbourne - said: "Alas, even an optimist would have to admit to extreme disappointment at the interim results of this study."
He pointed out that of the 343 schools originally approached to take part, 222 had not replied and 72 had said they were not interested.
This indicated the "extremely low priority" still given to sex education by schools.
Research showed that, in Britain, teenage boys had first intercourse with girls of roughly their own age. But three-quarters of girls were with an older man.
"Not only is this situation a perfect recipe for the spread of sexually-transmitted diseases, but it makes it doubly difficult for the young girl to insist on condom use," said Professor Short.
A lesson could be learned from the "Double Dutch" approach in the Netherlands, which has one of the lowest teenage pregnancy rates.
There, boys were encouraged to use condoms and girls to take oral contraceptives.
In Britain, taking the contraceptive pill off prescription, making it more readily available to teenagers, would be an "enormous breakthrough," he added.