The existing urban sewer databases in cities of developing countries present deficiencies in the records of the physical characteristics of the pipes, which limits their use as support for the implementation of management tools for said infrastructure. The objective of this work consisted in proposing and using in 13 sub-basins belonging to the El Salitre basin of Bogotá, comprising an area of 4515 Ha, with a total of 12,842 pluvial and combined pipes, decision trees and information entropy to identify pipes with demands for network maintenance activities such as cleaning, rehabilitation, replacement, etc., based on estimated flood reliability (for 2337 pipes whose physical characteristics were accurately recorded). The classificatory model presented an average predictive capacity of 62%. The pipes were classified using 28 rules, according to the reliability values, identifying 3383 pipes (26% of the total) with a high prioritization of proactive maintenance.