Objective: To validate ultrasound criteria for prediction of severe preeclampsia based on the generation of appropriate reference values of the uterine artery pulsatility index according to gestational age and maternal age in a large sample of Peruvian women. Material and methods: We performed a retrospective study of 8,392 pregnant women attended in a level III maternal perinatal referral center in Peru between 2009 and 2016. The sample was randomly selected and divided into 2 groups: the study sample, in which reference curves of the mean uterine artery pulsatility index were constructed according to gestational age and at-risk maternal age (≤15 years and >35 years); and the validation sample, in which the criteria were validated. Multivariate analysis was applied, and the diagnostic and predictive value of the criteria were measured. Results: The area under the ROC curve was 0.61 (95%CI, 0.58-0.64) for the mean uterine artery pulsatility index, 0.66 (95%CI, 0.63-0.68) for values higher than its 95th percentile, and 0.60 (95%CI, 0.57-0.63) for at-risk maternal age. The combination of a value greater than the 95th percentile and at-risk maternal age generated a greater significant area (0.72 [95%CI, 0.688-0.742]) than the others separately. Conclusions: The mean uterine artery pulsatility index helps us to predict severe preeclampsia. However, the application of values above the 95th percentile combined with at-risk maternal age improved the identification of severe preeclampsia in our population, with a specificity of 99% and a positive predictive value of 89%.