I mean, both cases, men to woman or woman to men.
No sarcasms, please, it’s only a discussion. 
Oops, sorry, I meant man to woman or woman to man.
Interesting question. I worked on this document for Health Canada, that synthesizes the current research on oral transmission of HIV. The chances are pretty low, but there is some risk. Only one study that we found calculated a per-sex-act probability of transmission for oral sex, of 0.04%, lower than any other form of intercourse (vaginal or anal). That study was on gay men, though, who have a higher rate of HIV prevalence than heterosexuals, so if you’re talking man-woman the chances are even lower. More likely it will be the person giving the action that will be more susceptible, but that doesn’t mean the receiver (the one getting the blowjob or otherwise being gone down on) is completely safe. Either way, nothing substitutes for knowing your partner’s history. And wearing a condom.
Just for the record, fellatio specifically refers to the act of oral sex by a man or woman on a man. The act from a man or woman on a woman is called cunnilingus.
I think Grousser means that if a woman performas fellatio on a man, what are the chances of transmission either way assuming one infected partner.
Depends what end you’re on.
Depends® is no protection against STD transfer. 
Strangely enough it comes from this website called www.straightdope.com. You should check it out sometime.
Is that assuming one partner is infected, or just the general probability of a random partner both having HIV and transmitting it during fellatio?
Cecil says there
I am very surprised by this assertion. Really? Using a condom correctly only reduces your chances by a factor of 10? I find this hard to believe. How can it be explained?
Does swallowing increase the risk for the party doing the fellating?
Only if the swallower has bleeding ulcers, or some other direct contact between their gastro-intestional tract and their bloodstream. And even then, it’d be a minor issue, since generally it’s blood leaking into the GI stream, not the other way around (because blood pressure is higher than GI tract pressure).
I recall a couple of studies that postulated that “deep throating” might increase the risk to the person giving head - I think the idea was that it could cause microtears to the lining of the throat, which would give HIV better access to the bloodstream. But this was very tentative and a number of years ago.
Once you’re at the low end of the risk scale, though, it all gets pretty speculative. Risk reporting seems to depend a lot on people’s memories of possible incidents in which they seroconverted. That’s not terribly reliable even assuming people are being truthful. And I suspect a lot of folks would only reluctantly admit that they were engaging in really risky sex.
Curiously, no one mentioned the cunilingus in the cafe discussion we had, but now that you mention it, include it please.
That was exactly the original question.
It’s curious, a lot of people says that oral sex is completely safe (even prostitutes, whom are supposed to be experts for their own safety). Whatever the sex of the “blowjob performer” was, it’s the common belief that the sperm doesn’t contains the HIV. So, there is no risk in swallowing.
What is really hard to determine is how to keep the pleasure for the “giver” with a condom. When trying, they often reacts with nausea.