Background: Protein coding gene annotation errors in prokaryotic genomes are accumulating continually, which decrease the quality of bioinformatics databases.In most cases, the update rate of genome annotation can not keep up with the explosive increasing number of completed genome sequences.Hence it is essential to manually rectify the genome annotation errors.However, the diverse re-annotation results obtained by different methods indicate that many existing re-annotation strategies are not credible enough for providing anticipative genome annotation tools.These differences also indicate potential increases of predicting false positive or false negative instead, which further accelerating the accumulation of genome annotation errors.Therefore, it is urgent to put forward more rigorous re-annotation strategy for genome annotations.