A few updates on recent studies....
oral sex transmission of HIV is a theoretical risk, but as to date, I do not know of any case reports of oral transmission. It is thought that saliva (which has antibodies in it) may neutralize the HIV before transmission can even happen.
The problem that has plagued a lot of the studies is that it's hard to find gay men who only practice oral sex where the HIV status of the other person is known. So, there's a lot of guesswork on probabilities.
There was a study published from a study of 981 subjects did find evidence that oral transmission may have been a factor. But compared to other activities like receptive anal sex, the probabilities were relatively low.
One study indicated that of the approximately 100 seroconversions approximately 8% of the subjects were infected through oral sex. Another study in San Francisco found no cases where oral sex could be the sole risk factor. Another study that looked a probabilities found that unprotected oral sex carries about 1/6 of the risk of unprotected anal sex.
What's more, HIV transmission is not absolutely guaranteed. The highest risk behavior (receptive unprotected anal) has a less than 1% transmission rate. If you are really worried and it's really been 3 years, get tested one more time, and then use condoms afterwards. Don't deprive yourself of the wonders of sex, just be safe.
Be very careful in citing these studies because they are studies of different populations in different risk categories.
There was a study done of heterosexual couples where one partner was HIV positive and the other was HIV negative (called discordant couples). It found that in couples who did not use condoms, the infection rate was about 10% (assumed that they were having vaginal and oral sex). Among the couples who used condoms, the infection rate was 0%.
There was another study looking at the risk from a single encounter in men who had sex with men where the bottom was HIV- and the top was HIV+. They calculated a risk from receptive anal sex without a condom to be 5 in 1,000 (0.5%) from a
single sexual act. This last clause is important because the risk increases with each act. So, if you have 2 encounters, the risk doubles to 10 in 1,000 (1%).
The bottom line on all these studies is that there is risk. Risk is associated with multiple partners, multiple encounters and high risk behaviors. If you're in a monogamous, long-term relationship with an HIV- person or if you limit the number of your partners and you practice safer sex, you will be in a low risk category.
It's a little like Russian Roulette- if I tell you that there is 1 bullet in one of the 6 chambers would that make a difference versus if I tell you there are 2 bullets in 6 chambers? Anyone with a rational mind wouldn't put a gun to their head.