Perhaps the most disturbing case of scientific fraud to come out in the last few years was made public in the NYT this week and is discussed in Bad Science. Dr Scott S Reuben has been found to have been fabricating his data for years.
What makes this so horrendous is not just the fact that he lied in 21 different studies, or that he actually made up his data, but that the potential costs of his lies are so high. If Dr. Reuben was studying something like the sexual behavior of butterflies, we probably wouldn't care too much. The problem is much more severe: his work was on the best way to manage pain after operations.
Add to all of this that he has been funded by Pfizer in recent years to study the effect of their pain medication drugs, and I am reminded yet again that current medical research practices are way behind where they should be.