Infants with possible cerebral palsy (CP) are commonly assumed to benefit from early diagnosis and early intervention, but substantial evidence for this is lacking. There is no consensus in the literature on a definition of 'early', but this review focuses on interventions initiated within the first 6 months after term age. We cover basic neuroscience, arguing for a beneficial effect of early intervention, and discuss why clinical research to support this convincingly is lacking. We argue that infants offered early intervention in future clinical studies must be identified carefully, and that the intervention should be focused on infants showing early signs of CP to determine an effect of treatment. Such signs may be efficiently detected by a combination of neuroimaging and the General Movements Assessment. We propose a research agenda directed at large-scale identification of infants showing early signs of CP and testing of high-intensity, early interventions.