[Retractions due to errors and frauds].

Scientific papers disseminate information but they cannot reflect the reality as they sometimes contain errors or fraud. Retractions and errata are the means allowing to correct the literature. The annual retraction rate increased 10 times between 1999 and 2009. Meanwhile, retractions are rare events, concerning 0.02% of the publications. To limit the errors and frauds, we should focus on quality instead of quantity of the literature; we should implement systems able to detect misconduct, and measures of prevention.