I bicycle a lot, but don't race. Oh, I've gotten a day license for a cyclo-cross race or two, but I'm much too slow to dream. With 30% below normal lung capacity due to sarcoidosis there's not much hope of competing with my age group.
Nevertheless, I follow racing a bit, and acquaintances ask me what I think happened with Floyd Landis. Here's my theory.
Only a tiny bit separates first from not-first at the top level of competition, and that difference is worth a lot of money. Plus, there are a lot of races, and so there's a lot of need to recover as fast as possible. Under these conditions, there are huge temptations to cheat. It's also hard and expensive to test. The testing protocols are statistical, and even the first level test for the A samples cost $300 each, so the testing is hardly continuous. At low levels of drug, the type that fly below the testing radar, you probably aren't doing much damage to your body (relative to what you are already doing by a career as a professional athlete).
So, temptation$ plus low likelihood of being caught plus little physical risk... that adds up to a high likelihood that there is a lot of cheating. There's evidence posted on this site http://www.cycling4all.com/index.php?content=d_news12.php that cycling and baseball have the highest rates of positive drug results.
Without any evidence whatever, I would guess that Landis was taking a complicated pattern of doping with testosterone (perhaps to improve recovery) and then adding epitestosterone as a masking agent to hide this (since the screener, $300 test, measures whether the ratio of testosterone to epitestosterone is above 4:1). In Landis's disappointment at finishing stage 16 so poorly, and perhaps because he had too much alcohol to drown his sorrows, he did not follow this regime to the letter.
The next day, he may have realized this might have been the case, which is why he drank so much water during the stage (a count of 70 water bottles either drunk or poured over his body was noted). This may have been an attempt to flush. In addition, a rider from the T-Mobile team drafted him during his entire ride, up until the final climb. The T-Mobile rider wasn't helping Floyd because his teammate Kloden was one of Floyd's rivals. The T-Mobile rider was trying to hang on to Floyd in hopes of having morestrength at the end and winning the stage in the final sprint. If so, Floyd would have had only a 1 in 70 chance of being tested (they test the winner, the leader, and two other random individuals) and would probably have gotten away with it. But the T-Mobile rider couldn't hang on and Floyd was tested.
So, my guess is Floyd was moderately doping, but messed up and got caught.