AbstractSome HIV vaccine candidates have a potential VEI (vaccine efficacy for infectiousness) type effect, which tends to reduce the viral load and may reduce infectiousness of an infected individual. In general, the efficacy of this kind of vaccine is very difficult to assess because it requires information on contacts of vaccinated infected individuals, and current methods to estimate VEI rely on the time elapsed between infections of an individual and his/her sexual partner, thus making infection of the sexual partner necessary to assess the efficacy. To avoid behavioural changes that may affect the estimates, HIV status is kept undisclosed to participants, which raises many ethical questions. Here we present a method that allows immediate notification of HIV status to both members of a couple, reducing the risk of infection when one of them has not been infected and allowing immediate medical treatment. The method allows for estimation of any VEI and VES (vaccine efficacy for susceptibility) effect, and it is robust to the most common situations found in these type of studies, namely: differential risk of participants, staggered enrollment and small sample sizes. Copyright © 2004 John Wiley & Sons, Ltd.