Cranial ultrasound abnormalities, especially a white-matter echolucency, predict disability (Paneth and Pinto-Martin 1990, Whitaker et al. 1990, Dammann and Leviton 1997), but with limitations. For example, half of very preterm infants who develop cerebral palsy (CP) do not have any abnormality of the white matter (Pinto-Martin et al. 1995, O'Shea et al. 1998). In addition, echolucencies that predict spastic CP are seldom located with descending fibers from the motor cortex (Rogers et al. 1994). These findings have led to the ‘tip of the iceberg’ hypothesis which postulates that cranial ultrasound scans do not detect the entirety of the white-matter damage (Leviton and Gilles 1996).