Introduction: Urinary tract infection (UTI) is the most frequent bacterial infection in infants with nonspecific clinical manifestations. Objective: To validate a predictive scale for UTI in febrile infants without apparent source and hospitalized in a neonatal unit. Patients and Method: A nested case-control study was conducted on 158 infants with febrile UTI, culture-confirmed, and on 346 febrile infants without apparent focus in whom UTI was ruled out, and also, hospitalized in a neonatal unit. The analysis was performed using Stata® 11. Associations were determined using odds ratio (OR) with 95% confidence interval. To find the predictive scale, multivariate analysis was performed using logistic regression and establishing major and minor criteria according to regression coefficient. Yield was calculated by sensitivity, specificity and area under ROC curve. The new predictive scale was validated by 108 new febrile neonates. Results: The major criteria to predict UTI were abnormal urinalysis and Gram positive cells without centrifugation, and among the minor criteria, male, age at time of fever, previous neonatal hospitalization, abnormal temperature (38.5°C or more, persistent fever, hypothermia) and 1.7 mg/dL C reactive protein or higher, resulting positive with the presence of one mayor or three minor criteria. It showed good performance with 100% sensitivity (CI 95%:98.3-100%), 92.3% specificity (CI 95%: 85.8-98.9%) and 0.962 area under ROC (95% CI: 0.932-0.991) when validated on 108 new febrile neonates without focus. Conclusions: The new predictive scale allows predicting UTI with good yield in infants with fever without an identified source.