Cerebral palsy is the commonest cause of severe childhood disability, the aetiology of which is largely unknown. Data on familial aggregation of cerebral palsy are very limited. We defined familial risks for siblings who were hospitalised because of cerebral palsy in Sweden. A nationwide database for neurological diseases was constructed by linking the Multigeneration Register to the Hospital Discharge Register for the years 1987-2001. Standardised hospitalisation ratios (SHRs) were calculated for affected singletons and twins by comparing them with siblings who had no cerebral palsy. A total of 3997 patients were recorded with cerebral palsy. Familial cerebral palsy was uncommon, and it accounted for 1.6% of all cerebral palsy cases. However, for parents who had had one affected child the risk of recurrence in another child was considerably increased. Parents of one affected child had a 4.8-fold risk of having a second affected child, and where the siblings were twins, the risk was 29-fold. These familial risks were particularly high in some clinical subgroups: 17-25 in singletons and 37-155 in twins, including hemiplegia, diplegia and quadriplegia. The remarkably high familial risks are difficult to explain without some contribution of heritable factors. The lack of discordant pairs may suggest that heritable factors are disorder type-specific. Affected concordant sibling pairs should be subjected to molecular studies aiming at identifying the susceptibility gene.