ABSTRACT Objective: To identify factors associated with the definition of the gestational age (GA) estimation method recorded in the live birth certificate (LBC), and to compare the results obtained according to the method in the city of São Paulo (CSP), between 2012 and 2019. Methods: Cross-sectional population-based study using the Live Birth Information System. Descriptive and comparative analysis was performed according to the GA estimation method, followed by a univariate and multivariate logistic regression model to identify the predictor variables of the method used. Results: The estimation of GA by the date of the last menstrual period (LMP) (39.9%) was lower than that obtained by other methods (OM) (60.1%) — physical examination and ultrasound, between 2012–2019. LMP registration in the LBC increased with the mother's age, it was higher among women who were white, more educated and with partners, in cesarean sections and with private funding. In the logistic regression, public funding was 2.33 times more likely than private funding to use OM. The proportion of preterm infants (<37 weeks) with GA by LMP was 26.5% higher than that obtained by OM. Median birth weight was higher among preterm infants with GA estimated by LMP. Conclusion: Prematurity was higher with the GA estimated by LMP in the CSP, which may indicate overestimation by this method. The source of funding was the most explanatory variable for defining the GA estimator method at the LBC. The results point to the need for caution when comparing the GA obtained by different methods.