The correct link does not have a zero at the end.
http://www.fda.gov/fdac/features/1997/babytabl.html
This site is discussing the effectiveness of condoms as birth control, so it is no wonder they come up with different numbers than sources discussing the effectiveness of condoms for HIV infection prevention.
This is one of the many problems with doing experiments. What is “Failure”? Once you have failure defined, you can make all the experiments you want and see what happens.
When condoms are cited to have a high failure rate (the 86% effective/14% failure) you say the FDA claimed) they are usually counting people who forgot or misused condoms as experimental failures.
When condoms are cited as having a very low failure rate (Esprix’s 2% failure rate/98% effective) the testers are typically looking at “perfect users”. Perfect Users are people who never forget, pay attention to expiration dates and always use water-based lube.
There is an interesting write up on condom effectiveness by consumer reports. I belive it is some years old know, but still interesting.
http://www.consumerreports.org/Special/ConsumerInterest/Reports/9906cnd1.htm
Searching for “FDA report effective condoms” brings up several good sites, including:
http://www.aids.wustl.edu/aids/condoms.html
This site has some good guidelines for how to use condoms properly.