The aim of this study was to review survival and certified causes of death of 584 children on a population-based register of cerebral palsy (CP), and to assess the impact on these of an additional diagnostic label. The register, established in 1985, comprised children with CP born from 1980 to about 1987 who were resident at the time of notification in the south-east of England in a region with between 3 and 4 million population, defined by the boundaries of the regional health authority (North East Thames Regional Health Authority). The current report concerns deaths of residents born between 1980 and 1986, who had been registered but had later died, as well as of eligible children who had not been registered while alive but whose cause of death was CP. These were identified at the Office of Population Censuses and Surveys who also supplied copies of death certificates for this study. For children notified while alive, information about motor severity and other diagnostic labels was sought at entry to the register and again between 3.5 and 4 years and between 7 and 8 years. For this study, children known to have a postneonatal onset, a progressive or non-cerebral cause of motor signs, or minimal motor involvement were excluded. Thirty-nine of 584 children included in this study had died by the end of 1995. No deaths had occurred in children known to have less than four-limb involvement. Survival of the group known to have an additional diagnostic label was significantly lower (86.2%) than that of the group with no known label (96.3%; P=0.01), and remained lower, although not significantly, if only those with severe four-limb involvement were compared (75.3% versus 92.4%; P=0.2). The greater severity of limb involvement in those with an additional diagnosis may not account for this difference. Of the 37 children with death certificates available, CP was mentioned as a cause in only 24.