Abstract: NaÏve Bayes algorithm has high accuracy in solving the spam classification field and can distinguish the mail very well. However, in the pre-classification training phase, it consumes a lot of system and network resources and seriously affects the classification efficiency. Spark platform for this introduction. With parallel thinking to solve the problem of mail classification, the use of spark computing platform RDD kinship rational arrangement of NB mail classification of the various processes. The experimental results show that, compared with other traditional classification methods, Na?ve Bayes has a good effect on the Precision and Recall rate, etc., and with the traditional mail classification under single machine, this experiment because of the introduction of distributed thinking, The use of spark clusters greatly accelerate the classification speed.