Hydrological studies for sizing urban drainage systems in the Amazon have often been neglected and little investigated for rainwater projects. This research evaluated alternative hydrological models used in sizing urban drainage network projects in subdivisions with subsidized houses in the Amazonian region in Brazil. Statistical tests of these models were performed for both original and alternative scenarios. The methodological steps we conducted as follows: 1) evaluate the dimensioning of infrastructure project networks, considering two case studies contemplated by the Calha Norte Program (CNP) in the state of Amapá; 2) test the statistical significance of the dimensioning of network diameters (α < 0.05), considering a) benchmark project (MD or M1) approved by the Ministry of Defense; b) determination of concentration time (C t ) and rainfall intensity-duration-frequency (IDF) relationships, as well as estimating diameters using alternative models. The results indicated a significant influence on the diameters of the projected rainfall networks (p < 0.05), suggesting that alternative models predicted more unfavorable flow peaks than the original model. We conclude that the benchmarking model underestimated the diameter of the project compared to alternative models, which means the optimized C t parameter significantly impacts dimensioning estimates in rainwater projects in these Amazonian municipalities. This suggests that underestimated parameters in MD may cause inefficiency in the stormwater system projects in future similar scenarios.