Back to Bareback

Does Risk Compensation Actually Happen?

Does Risk Compensation Actually Happen?
Bareback Guide 12 February 2026

Risk compensation is the idea that people offered protection respond by taking more risk, offsetting some of the benefit. It was the central objection to PrEP before it was rolled out, and the evidence since has been more mixed than either the enthusiasts or the critics predicted.

What did the trials find?

No increase in HIV among people taking PrEP, which is the outcome that matters most. Increases in bacterial sexually transmitted infections were observed in some studies and not others, and where they occurred they were modest. The UK PROUD study found high rates of bacterial infections in both arms, suggesting the population was already at high risk rather than becoming so.

What has happened since rollout?

Condom use among gay and bisexual men has continued to decline, and bacterial infection rates have risen. Attributing this to PrEP is complicated by two things: condom use was already declining before PrEP arrived, and PrEP brought large numbers of people into three-monthly screening at every site, which finds silent infections that were previously never counted.

How much of the rise is real?

Genuinely difficult to separate. Rectal and throat infections are silent in the great majority of cases, so a population that starts being screened quarterly will appear to have more infections regardless of whether anything changed. Both a real behavioural shift and an ascertainment effect are almost certainly contributing, and the proportions are debated.

Is compensation irrational?

Not obviously. Someone who was using condoms primarily because of HIV, and who has now removed HIV risk by another means, is behaving coherently by using them less. The framing of risk compensation as a failure of discipline assumes condoms were an end in themselves rather than a tool for a purpose. If the purpose is served differently, the tool becomes optional.

Where does the argument go wrong?

When it is used to argue against providing prevention. The logic that a protective measure should be withheld because people might respond by taking more risk has been applied to seatbelts, contraception and needle exchanges, and has been wrong in each case. What matters is the net outcome, and for HIV the net outcome has been a substantial fall in diagnoses.

What is the legitimate concern?

That bacterial infections carry real consequences, that gonorrhoea is becoming harder to treat, and that a prevention strategy focused entirely on HIV leaves those unaddressed. This is a reasonable point and it argues for adding vaccination, screening and doxy-PEP rather than for restricting PrEP.

Does it apply to other interventions?

The same question is raised about doxy-PEP and about U equals U. The evidence in each case is limited and the reasoning is the same: the intervention works, the question is what else needs to be in place alongside it.

What should you take from the debate?

That PrEP does what it claims and covers nothing else, which is not compensation but a fact about the drug. If you have reduced condom use since starting PrEP, the useful response is to check that vaccination, screening and doxy-PEP are covering what condoms were, rather than to feel judged about a decision that was reasonable.

The honest summary

Some behavioural change has occurred, its size is uncertain, and it does not undermine the case for PrEP. It does strengthen the case for everything that sits alongside it.

Does it apply to individuals or populations?

The research describes populations, and applying it to an individual is a mistake. Plenty of people take PrEP and use condoms exactly as before; plenty stopped using condoms before PrEP existed. A population-level shift says nothing about what any particular person is doing.

What would change the picture?

Better data separating genuine behavioural change from increased detection, which requires knowing what the screening rate was before. Some studies have attempted this and the results are mixed. It is an open question rather than a settled one in either direction.