The risk "is low but non-zero," they warn, adding that if people seized upon the Swiss statement as guidance, incidence of HIV among couples could easily quadruple over a decade.
Their statistical analysis appears in the British medical weekly The Lancet.
In January, Switzerland's Federal Commission for HIV/AIDS sent ripples across the tight-knit community of AIDS experts and activists in a report on so-called "discordant" couples.
These are couples in which one individual has HIV while the other does not.
The panel said that such couples need not use a condom provided the infected partner regularly followed antiretroviral therapy and had no genital infections.
In addition, the level of human immunodeficiency virus (HIV) in his or her blood had to be below detectable levels -- a threshold of 40 viruses per millilitre of blood -- for the six previous months at least.
Its statement was condemned by activists as irresponsible, while experts criticised the scientific basis for making it.
They accused the evidence of being sketchy and founded on heterosexual couples, rather than gay couples, and on vaginal intercourse, rather than anal sex, where the infection risk is much higher.
In The Lancet paper, epidemiologists led by David Wilson of the University of New South Wales in Sydney used a simple mathematical model to estimate the cumulative risk of transmission if the infected partner had only 10 viruses per millilitre of blood.
Their model is based on the assumption that each couple has 100 unprotected sexual encounters a year.
If so, an HIV-negative male would face a 0.22-percent risk of infection each year from an HIV-positive woman. The risk for an uninfected female from an infected male would be 0.43-percent annually.
The risk in gay relationships would be far higher -- in male-to-male anal transmission the probability of infection would be 4.3 percent per year.
Extrapolated over a decade, this would translate into a fourfold increase in AIDS infections among discordant couples, compared with current rates of condom use.
"If the claim of non-infectiousness in effectively treated patients was widely accepted, and condom use subsequently declined, then there is the potential for substantial increases in HIV incidence," the study warns.
The antiretroviral drug "cocktail," introduced in 1996, has been a lifesaver for many people with HIV.
But it is not a cure, for it only suppresses viral levels, rather than eradicates the pathogen completely. If the drugs are stopped, the virus rebounds.