BACKGROUND: Shoulder dystocia is an obstetrical emergency that requires immediate and appropriate provider response to decrease neonatal morbidity. OBJECTIVE: To determine whether institution of shoulder dystocia simulation training improves maternal and neonatal outcomes at delivery. STUDY DESIGN: This retrospective cohort study compared maternal and neonatal outcomes before (1 September 2008-31 December 2009) and after (1 January 2010-31 December 2014) a mandatory shoulder dystocia simulation curriculum was instituted. Χ(2) and Fisher's exact tests along with multivariate logistic regression models were conducted to adjust for potential confounding. RESULTS: Of 8930 vaginal deliveries that met eligibility criteria over the 6-year period, 299 (3.3%) deliveries were complicated by a shoulder dystocia. The adjusted frequency of shoulder dystocia was approximately two times higher after training implementation (1.9% vs 3.8%; adjusted OR=1.80 (1.23 to 2.65), p=0.003). Neonatal outcomes such as brachial plexus injury (5.0% vs 7.7%; p=0.75), clavicle fracture (5.0% vs 4.6%; p=1.00) and humerus fracture (0.0% vs 1.9%; p=1.00) were not statistically different after training. Additionally, no significant differences in episiotomy (5.0% vs 4.6%; p=1.00), postpartum haemorrhage (10.0% vs 12.4%; p=0.80) and severe perineal laceration (10.0% vs 6.6%; p=0.50) were observed. CONCLUSIONS: The overall frequency of shoulder dystocia reportedly doubled after training despite stable operative vaginal delivery and caesarean delivery rates. No significant improvements were seen in maternal or neonatal outcomes after simulation training. As shoulder dystocias are rare events, simulation training may not necessarily improve neonatal outcomes, but it may increase overall provider awareness and intervention.