This work aims at proposing a rainfall classification criterion for the city of Manaus based on the adoption of clustering algorithms. In order to achieve such result, it was considered data from 2010 to 2016, adoption of 5 different clusters of events and the use of k-means and k-medoids algorithms. The results obtained with k-means, in particular, were more effective in capturing the higher frequency of low intensity rainfall events as well as the lower frequency of extreme rainfall events. The results obtained can be adoted as rainfall classification criterion for Manaus.