The most burning criticism, levied by American epidemiologist James Chin in his book "The AIDS Pandemic," accuses UNAIDS of intentionally inflating its estimates of how many people have HIV in order to dramatise the epidemic and win more money from donors.
Chin appeared vindicated in June, when India dramatically reduced its estimate of people infected to a range of 2.0-3.1 million from 5.7 million.
The earlier estimate had given India the tragic distinction of having more cases of HIV than any country in the world, including South Africa -- even though the virus affects a far greater percentage of South Africans.
International health experts have for years worried about the possibility of an AIDS epidemic in India on the scale of South Africa's, but the new figures suggested a fairly low infection rate of 0.36 percent of the adult population.
The infection rate was earlier presumed to be 0.9 percent.
An equally sharp reduction was made in Cambodia, where the estimated infection rate was cut to 0.6 percent of adults from 1.6 percent.
UNAIDS regional director for Asia Prasada Rao told AFP the decreases were only about improved science and said the dramatic reduction in the estimates showed the agency's willingness to embrace new data.
"Ascribing motives to UNAIDS, that you are deliberately inflating the numbers to get more resources, that's really not something which I believe is acceptable," Rao said.
In both India and Cambodia, Rao said the reduced estimates were made because of improved surveys.
Earlier estimates were made based on data from clinics around each country, which experts used to guess how many people in the entire population have HIV.
The new data came from random surveys of households, which provides a broader picture, he said.
"I don't see any motive on the part of UNAIDS to inflate numbers. I don't think there is any axe to grind in this case," Rao said.
Whichever method is used to make the estimates, scientists agree the numbers remain a guess best used to show trends in each country.
But Chin said he believes UNAIDS had intentionally used the upper end of all its estimates to try to make the epidemic seem as devastating as possible.
"There is a fine line between deliberately lying with the numbers or using the upper range of estimates that are based on slim assumptions and unrepresentative data," he told AFP in an email.
Both men say the debate isn't really about numbers but about deciding how to spend often scant public health resources.
Helen Epstein, who has spent years studying AIDS in Africa, argues that the UN agency has missed the mark on that point as well.
In her book "The Invisible Cure" she says UNAIDS either misunderstood or ignored data in the mid-1990s that showed many people in eastern and southern Africa had multiple long-term sexual partners, creating a web of relationships that allowed HIV to spread quickly.
"Vast numbers of people were caught up in a network of sexual relationships that they didn't realise were so risky," she said.
Instead of trying to encourage people to stop having more than one partner, as Uganda successfully did with a "Zero Grazing" campaign in the 1980s, Epstein says AIDS activists focused on condoms and abstinence which proved less effective but were politically more appealing to liberals and conservatives.
The patterns of the disease in Africa are different to those in Asia, for example, where it spreads mainly through high-risk groups like prostitutes, intravenous drug users, and men who have sex with men, she said.
"The thing that bothered me is the way that they overblew the prospects for an Asian epidemic," she said. "They got it almost perfectly wrong in some places. They pushed for a targeting of the general population in places like India, when they probably should have targeted high risk groups."
But in Africa, the high number of people infected "includes absolutely everybody -- teachers, doctors, farmers, market traders, politicians, everyone".
Epstein said UNAIDS has tended to cite the prospect of the disease exploding in other parts of the world, without focusing enough on African nations that have long been the hardest-hit.
The books have sparked a wide-ranging debate among activists about how to fight the disease and raised questions about UNAIDS leadership.
Rao insists the agency is open to criticism and has been adapting as new data becomes available.
He worries that if UNAIDS is seen to be manipulating information, that could damage the agency's reputation and ultimately efforts to stop the disease.
"UNAIDS is not saying the data is wrong. It is accepting the data and trying to harmonise the facts," he said.
"That shows the openness that the organisation has got on this issue. And it is prepared to correct its data (and) revise its data based on other sources of information."