OBJECTIVES To report epidemiological trends in cerebral palsy including analyses by severity.

DESIGN Descriptive longitudinal study in north-east England. Every child with suspected cerebral palsy was examined by a developmental paediatrician to confirm the diagnosis. Severity of impact of disability was derived from a parent completed questionnaire already developed and validated for this purpose.

SUBJECTS All children with cerebral palsy, not associated with any known postneonatal insult, born 1964–1993 to mothers resident at the time of birth in the study area.

MAIN OUTCOME MEASURES Cerebral palsy rates by year, birth weight, and severity. Severity of 30% and above defines the more reliably ascertained cases; children who died before assessment at around 6 years of age are included in the most severe group (70% and above).

RESULTS 584 cases of cerebral palsy were ascertained, yielding a rate that rose from 1.68 per 1000 neonatal survivors during 1964–1968 to 2.45 during 1989–1993 (rise = 0.77; 95% confidence interval 0.2–1.3). For the more reliably ascertained cases there was a twofold increase in rate from 0.98 to 1.96 (rise = 0.98; 95% confidence interval 0.5–1.4). By birth weight, increases in rates were from 29.8 to 74.2 per 1000 neonatal survivors < 1500 g and from 3.9 to 11.5 for those 1500–2499 g. Newborns < 2500 g now contribute one half of all cases of cerebral palsy and just over half of the most severe cases, whereas in the first decade of this study they contributed one third of all cases and only one sixth of the most severe (χ² and χ² for trend p < 0.001).

CONCLUSIONS The rate of cerebral palsy has risen in spite of falling perinatal and neonatal mortality rates, a rise that is even more pronounced when the mildest and least reliably ascertained are excluded. The effect of modern care seems to be that many babies < 2500 g who would have died in the perinatal period now survive with severe cerebral palsy. A global measure of severity should be included in registers of cerebral palsy to determine a minimum threshold for international comparisons of rates, and to monitor changes in the distribution of severity.