AIM This study aims to identify characteristics at 2 years of age that differ between children with confirmed cerebral palsy (CP) and a non-CP diagnosis by 5 years of age.METHOD This was a retrospective cohort analysis. A CP diagnosis may be considered a 'probable' diagnosis at 2 years, which is often 'confirmed' at 4 or 5 years, particularly in the context of CP registries. A total of 1683 children with a diagnosis of CP or probable CP at 2 years of age were identified from the Canadian Cerebral Palsy Registry, of whom 48 received a non-CP diagnosis at 5 years ('non-confirmed CP'). Perinatal adversity, preterm birth status, Gross Motor Function Classification System (GMFCS) level, presence of comorbidities, magnetic resonance imaging (MRI) findings, and initial CP motor type were compared between the two groups by univariate and logistic regression analyses.RESULTS v 2 analysis and multivariate analysis both confirmed that children with a non-CP diagnosis by 5 years of age were more likely to have a normal MRI (v 2 odds ratio [OR]=7.8, 95% confidence interval [CI]=3.8-16.1; OR=5.4, 95% CI=2.4-12.5), ataxic-hypotonic (v 2 OR=10.1, 95% CI=4.9-21.2; OR=6.1, 95% CI=2.2-16.2) or dyskinetic CP (v 2 OR=2.7, 95% CI=1.2-5.9; OR=2.9, 95% CI=1.0-7.6), born at term (v 2 OR=3.7, 95% CI=1.7-8.0; OR=3.6, 95% CI=1.0-12.1), and lack perinatal adversity (v 2 OR=4.1, 95% CI=1.6-10.7; OR=3.4, 95% CI=1.0-11.7).INTERPRETATION Normal MRI, ataxic-hypotonic or dyskinetic CP, lack of perinatal adversity, and term birth are associated with a higher odds of non-CP diagnosis by 5 years of age, thus potentially enhancing diagnostic work-up.