BACKGROUND: The assessment of the newborn after birth is fundamental for identifying those requiring resuscitation. Certain components of the Apgar score are used to assess neonatal condition, but their value is insufficiently validated. We aimed to identify the components of the Apgar score that predict neonatal resuscitation. METHODS: Individual patient data from two multicenter trials were analyzed. Preterm newborns under 32 weeks of gestation and term newborns with perinatal acidosis and/or resuscitation were included. The extent of resuscitation was quantified by a standardized scoring system, and the clinical condition was quantified by a specified Apgar score. Correlation, linear regression and ROC analyses were used to address the study question. RESULTS: A total of 2093 newborns were included. Newborns in poor clinical condition at 1 min received more interventions at 5 and 10 min. Heart rate, muscle tone, reflexes and breathing quality, but not skin color, were moderately strong correlated with the extent of resuscitation at 5 (r = 0.44 to 0.52) and 10 min (r = 0.38 to 0.42). Heart rate, reflexes and chest movement at 1 min best predicted the subsequent need for resuscitation (R CONCLUSIONS: The clinical impression at 1 min can help to predict the need for medical interventions. Contrary to recent guidelines, heart rate, reflexes and chest movement seem to have the highest values. TRIAL REGISTRATION: The Test APGAR study was registered at clinicaltrials.gov (NCT00623038, 14/08/2008).