We compared rates of preterm and postterm birth according to six algorithms for gestational age (GA) estimates based on last menstrual period (LMP) and early ultrasound (EUS): LMP alone, LMP if the discrepancy between the two estimates was within 14 days and otherwise EUS (14-day rule), a 10-day rule, a seven-day rule, a three-day rule and EUS alone. In a sample of 44,623 births in a Canadian tertiary hospital, the choice of algorithms makes a substantial impact on both preterm and postterm birth rates, even when EUS was used for discrepancies over two weeks.