Information provided by the neonatal neurologic assessment is important for identifying infants with neurologic abnormalities at a very early age. The aim of this study was to compare two distinct approaches to the neurologic assessment of newborns: the Amiel-Tison neurologic assessment, and Prechtl's qualitative assessment of general movements. The results of both assessments were studied in a group of 45 preterm infants with different risk factors for brain damage that were compared at term age and at a corrected age of 3 months. The predictive power of the two methods regarding the neurologic and developmental outcome at a corrected age of 12-15 months was analyzed. The agreement of the two methods was excellent at term (kappa = 0.87) and good at 3 months (kappa = 0.54). The sensitivity of both methods for detecting children with neurologic abnormalities was high both at term and at 3 months (0.92, 1.0 for the Amiel-Tison assessment and 0.96, 1.0 for general movements). The specificity of both techniques was low at both ages (0.45, 0.75 for the Amiel-Tison assessment and 0.40, 0.35 for general movements). The agreement of the neurologic and developmental outcome was better with the Amiel-Tison assessment (kappa = 0.39, 0.77) than with the observation of general movements (kappa = 0.38, 0.37).