The report was published Friday in the British medical journal, Lancet.
Levonorgestrel, the most widely used emergency contraceptive pill, is only effective if women take it within three days of having sex. It is sold under various brand names including Levonelle and Plan B, and is available in more than 140 countries, including the United States, Canada and many countries in western Europe. In nearly 50 of those countries women can get it without a prescription.
International researchers compared Plan B to the new drug ulipristal acetate, sold as ellaOne in Europe only with a doctor's prescription. The drug is not legally on the market elsewhere.
Experts tracked nearly 1,700 women aged 16 to 36 who received emergency contraception within three to five days of having unprotected sex. About half got Plan B while the rest got ellaOne.
In the group that got Plan B, there were 22 pregnancies. In those that got ellaOne, there were 15. In both groups, the most frequently reported side effect was a headache. The research was paid for ellaOne's maker, HRA Pharma, which helped design the study.
When the researchers pooled their results with a previous study comparing the two morning-after pills, they found women who took ellaOne within five days after sex almost halved their chances of becoming pregnant compared to women who took Plan B.
Women who took ellaOne had a 1.8 percent chance of becoming pregnant, while women who took Plan B had a 2.6 percent chance.
Last May, the European drug regulator approved ellaOne. HRA Pharmacy has refused to say where else the pill is being submitted for approval.
Compared with Plan B, which becomes less effective over time, ellaOne appears to work consistently well over five days in women who have unprotected sex.
Health officials, however, warned that this should not give women a false sense of security.
"The message has to be always that women should act as soon as possible," said Tony Kerridge, a spokesman for Marie Stopes International, a nonprofit sexual health organization in Britain not linked to the study.