Objective To determine whether the ability to predict severe motor impairment at age 5 years improves between birth and 18 months. Design Ancillary study of the Caffeine for Apnea of Prematurity Trial. Setting and Patients International cohort of very low birth weight children who were assessed sequentially from birth to 5 years. Outcome Measures Severe motor impairment was defined as a score <5th percentile on the Movement Assessment Battery of Children (MABC), or inability to complete the MABC because of cerebral palsy. Multivariable logistic regression cumulative risk models used four sets of predictor variables: early neonatal risk factors, risk factors at 36 weeks' postmenstrual age, risk factors at a corrected age of 18 months, and sociodemographic variables. A receiver operating characteristic curve (ROC) was generated for each model, and the four ROC curves were compared to determine if the addition of the new set of predictors significantly increased the area under the curve (AUC). Results Of 1469 children, 291 (19.8%) had a severe motor impairment at 5 years. The AUC increased from 0.650 soon after birth, to 0.718 (p<0.001) at 36 weeks' postmenstrual age, and to 0.797 at 18 months ( p<0.001). Sociodemographic variables did not significantly improve the AUC (AUC=0.806; p=0.07). Conclusions Prediction of severe motor impairment at 5 years of age using a cumulative risk model improves significantly from birth to 18 months of age in children with birth weights between 500 g and 1250 g.