What the science says about condoms

What the studies say overall

The condom is among the most studied tools in sexual health. A Cochrane review of heterosexual couples in which one partner had HIV estimated that always using one reduces new infections by around eighty percent (Weller et al., 2001).

And using the female condom together with the male condom prevented more sexually transmitted infections than using the male condom alone, according to a review of fifteen studies (Wiyeh et al., 2020).

Non-latex condoms, such as polyurethane ones, are the alternative for people with a latex allergy. A meta-analysis found that they break more often than latex ones, although they remain an acceptable option (Gallo et al., 2006).

The trials do not fully agree: in one, breakage of the polyurethane condom was no different from latex, in another it had more failures although all were low, and in a third it was not shown to protect equally well against pregnancy (Potter et al., 2003; Potter et al., 2024; Steiner et al., 2003).

Many failures have to do with how the condom is used. In a trial of almost five thousand acts of intercourse, using lubricant reduced breakage and slippage, in both anal and vaginal sex (Siegler et al., 2019).

Among university students in the United States, fit problems and other user errors were associated with breakage, and in Flanders discomfort and fit appear as barriers to using one (Butler et al., 2025; Dewaele et al., 2024).

In two studies of men who have sex with men, failures were common, and more so under the influence of substances (Sandfort et al., 2025; Gray et al., 2024). Talking about condoms with a partner was associated with using them consistently (Edison et al., 2022).

The type of lubricant matters: in a laboratory study from 1989, one minute of contact with mineral oil reduced the strength of latex by around ninety percent (Voeller et al., 1989).

The female condom, for its part, takes practice: among women in Durban, failures dropped sharply between the first and fourth period of use, and half of a group of Australian women had some difficulty inserting it (Beksinska et al., 2015; Fenwick et al., 2021). The newer designs did not fail more than the reference model (Beksinska et al., 2013).

Where they disagree

On the material, the trials do not give a single result: polyurethane fails somewhat more than latex in some studies and the same in others, and for preventing pregnancy one did not show the same effectiveness (Gallo et al., 2006; Potter et al., 2003; Potter et al., 2024; Steiner et al., 2003).

Breakage figures also vary a great deal depending on the population and the situation: a stable couple in a trial is not the same as intercourse under the influence of substances (Sandfort et al., 2025; Gray et al., 2024).

Limits of the evidence

The large figure for protection against HIV comes from older studies of heterosexual couples (Weller et al., 2001). Several studies are surveys that depend on what people remember, and the mineral oil one is a laboratory test, without people (Butler et al., 2025; Dewaele et al., 2024; Voeller et al., 1989).

None compares specific brands.

The studies, one by one

We have gathered 16 verified studies. Each summary is our own; the quotation in inverted commas is taken word for word from the original abstract, in English.

Other topics in the library

• Vaginal lubricants13 studies
• Anal sex12 studies